From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Authentication-Results: plum.tunbury.org; dkim=pass (1024-bit key; unprotected) header.d=inria.fr header.i=@inria.fr header.a=rsa-sha256 header.s=dc header.b=fCeX9Krn; dkim=fail reason="signature verification failed" (1024-bit key; secure) header.d=polytechnique.org header.i=@polytechnique.org header.a=rsa-sha256 header.s=svoboda header.b=WNsKfbZx; dkim-atps=neutral Received-SPF: Pass (mailfrom) identity=mailfrom; client-ip=192.134.164.83; helo=mail2-relais-roc.national.inria.fr; envelope-from=caml-list-owner@inria.fr; receiver=tunbury.org Received: from mail2-relais-roc.national.inria.fr (mail2-relais-roc.national.inria.fr [192.134.164.83]) (using TLSv1.2 with cipher ECDHE-ECDSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by plum.tunbury.org (Postfix) with ESMTPS id 09E14400A2 for ; Tue, 8 Jul 2025 12:45:46 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=inria.fr; s=dc; h=from:to:date:message-id:mime-version:subject:reply-to: sender:list-id:list-help:list-subscribe:list-unsubscribe: list-post:list-owner:list-archive; bh=CDQ6shqn2fn/x8zjx29cyg0dbKsiAwoGOE+Hbte8+zg=; b=fCeX9Krn179csau+EsAi/2omwKFCa9tXfk6k+9pni6by4TPNyIvkyAZj WTPK1X/ha6xVM8OD9ylL2UpqLcGvPXM/zGUg3kwNkixhOZA8wuwJnBS5U LsG5S93YmwkurqkfXwMa99MeNG28CEXij9WiuGVnYbxLCs4CFwg5nd1CA o=; Received-SPF: Pass (mail2-relais-roc.national.inria.fr: domain of caml-list-owner@inria.fr designates 128.93.162.160 as permitted sender) identity=mailfrom; client-ip=128.93.162.160; receiver=mail2-relais-roc.national.inria.fr; envelope-from="caml-list-owner@inria.fr"; x-sender="caml-list-owner@inria.fr"; x-conformance=spf_only; x-record-type="v=spf1"; x-record-text="v=spf1 include:mailout.safebrands.com a:basic-mail.safebrands.com a:basic-mail01.safebrands.com a:basic-mail02.safebrands.com ip4:128.93.142.0/24 ip4:192.134.164.0/24 ip4:128.93.162.160 ip4:128.93.162.3 ip4:128.93.162.88 ip4:89.107.174.7 mx ~all" Received-SPF: None (mail2-relais-roc.national.inria.fr: no sender authenticity information available from domain of postmaster@sympa.inria.fr) identity=helo; client-ip=128.93.162.160; receiver=mail2-relais-roc.national.inria.fr; envelope-from="caml-list-owner@inria.fr"; x-sender="postmaster@sympa.inria.fr"; x-conformance=spf_only Authentication-Results: mail2-relais-roc.national.inria.fr; spf=Pass smtp.mailfrom=caml-list-owner@inria.fr; spf=None smtp.helo=postmaster@sympa.inria.fr; dkim=hardfail (body hash did not verify [final]) header.i=@polytechnique.org X-IronPort-AV: E=Sophos;i="6.16,297,1744063200"; d="scan'208,217";a="230817275" Received: from prod-listesu18.inria.fr (HELO sympa.inria.fr) ([128.93.162.160]) by mail2-relais-roc.national.inria.fr with ESMTP; 08 Jul 2025 14:45:44 +0200 Received: by sympa.inria.fr (Postfix, from userid 20132) id D8E58E0AF6; Tue, 8 Jul 2025 14:45:43 +0200 (CEST) Received: from mail3-relais-sop.national.inria.fr (mail3-relais-sop.national.inria.fr [192.134.164.104]) by sympa.inria.fr (Postfix) with ESMTPS id 339DCE0157 for ; Tue, 8 Jul 2025 14:45:42 +0200 (CEST) IronPort-SDR: 686d12f3_VAplU53yArwVJzlB2XlJ4bWcjDuWTwJ1FE7gUCFqUnOfRjU 33cqEC6wrvPBxcpNAi7z3ZD7XehRXdRkAeNISzw== X-IPAS-Result: =?us-ascii?q?A0EvBQCSEm1odyIeaIFSCIQdWygZAWAGWjMHCEkDhBY8g?= =?us-ascii?q?WOBbI4igRaKUYVkiEqCMIFpgUIjFQEDAQ0uARsEAQIEAQEDAQIBggyBPYE3A?= =?us-ascii?q?ot8Ah8GAQQ0EwECBAEBAQEDAgMBAQEBAQEQAQEFAQEBAgEBAgQGAQIQAUNJh?= =?us-ascii?q?XsNgkUZOEEwdAEBAQEBAQEBAScBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBA?= =?us-ascii?q?QEBAQEBAQEBAQEBAQEBAQEBAgwBAoEADg4BCAQGEwEBKQEBAQMDBhgUDwMQB?= =?us-ascii?q?AEGAwIRARAlAwETARIUBoJpgh8CAhYDMgMEAQwGkw2aeho3en8zgQFXgTUBA?= =?us-ascii?q?QaBCD4DCwICAw8uAdkQDW6BG0oJgUmFbYJGBBoBKkhrAoRICYQzJw+BVUSBF?= =?us-ascii?q?YIoSgdvgUJdIAsXAQEBAVxEBAQBAQ0EAgEaDyQJgyWCaYIRFUQ+FB2BDYESg?= =?us-ascii?q?nKBQGESgSpDghoEgVyEO4ZmgUQiAyYzLAFVExcLBwVbgQgDKjQxbjIdgSeGe?= =?us-ascii?q?4QqK0+FEIEZg1QRE20PBoEdgSJAAwttPTcUG5JgGSkacAIBgUklDjgGKwMED?= =?us-ascii?q?A8NCAIRAQYOBQEFDhAGAiACDSUDASMWBwMFDwEEDwIHAQ0FCRUBBiMFBgkCC?= =?us-ascii?q?wItA5JBCAwlAwYkAgF0jWl4oW8dMD00B4QfgV4GDIkIgSaOMoEKgnKDV4QET?= =?us-ascii?q?YEKiziHApJPIphkIoI2hycKgRwJgW9qhAiRVQIRAwYEhTmBfyM8MHAzGjBDg?= =?us-ascii?q?mcJRhwPiACFfy4WgRIBAQIEAYJDQX2BJnopUTuCZrcMQTUCAQEKLgIHAQoBA?= =?us-ascii?q?QMJhUUdAQGIDYFDCAEB?= IronPort-PHdr: A9a23:XlO2IxWs4Ax7MEQNKt9sidvQOKbV8KxlWjF92vMcY1JmTK2v8tzYM VDF4r011RmVBt6ds6kP0LaJ6ujJYi8p39WoiDM4TNR0TRgLiMEbzUQLIfWuLgnFFsPsdDEwB 89YVVVorDmROElRH9viNRWJ+iXhpTEdFQ/iOgVrO+/7BpDdj9it1+C15pbffxhEiCCybL58I hi6txndu8cZjYZsJas8yAbFqWZUdupLwm9lOV2ckxHg68mq+5Jt7zpesO87+c5aVqX6caU4T bhGAzkjLms4+s7luwTdQAWW/ncTXXkYnRROAwje8RH1RYzxvTfgtup8wyaVI8v7Rq0pVDu47 qdrTBjoiDobNzM87WrahNB8gL5drRm8pxBxwIjUYJ+UNPplf6PSZtcaSnRcVcdfUCxBGJ6zb 5ASBOYFM+tUs4zxql0TphW8GAasHvvixD9WiHHr06M00OsuHh3d0QM6A94Dqm7ZoMnpOKoQV +2+0anGzS/Eb/NTwTrw9IzIfQonofqRQLx/bcrRyUwuFwPZklWbtIvoMCmR1usXtWiX9fFgW v6vi24iqgFxviSvy9w0ionOgoIUykzE+jtlz4Y1I924SVd7YcO6H5dKsCGaLYR2T9okTmp1t yk01qcItoSnfCgW1psn3RjfZuSHfoaG7B/tVOScLDZkiX9qeLyyhhS//0avxODiSMS50ldHo yVBn9TDsn0AywHe58eJRPV9/0qsxCuC2gTQ5+xKJ00/iKTVK5kkwrEql5oTt1zOHjfol0Xqj a+Walsr+vCv6+TiZ7XpuIWQN4lqhQHiKqgundCwDv49MggKRWSb/v681LL78U32QbVKkv02k rTCv5zAOcsboau5Dxda0oYi9xa/Dyqm388CkXUdMF1FfxeHg5DpO17UO/D4Dumwg1CokDtxw PDGJLLhDo3QIXjEjLjhfqhy61RGxAUvytBf4opYCrAHIPLoRk/+rtjYDhsjPwyvw+brEMly1 oMEVmKJDa+WLrjSvUWN5u0yO+WMfI8UuCr6K/g/+/HujWU1mVgHfammxZcXcmq3Hup4LEWfe Hbsh9MBHX0XsQUgUuzmkEONUTBdZ3msQ608+ig3CIK8AofFW4+thKKO3D2gHpFMYWBGEF+MH W/yeIWAVPcMbziSIs59kjwEULihSpMu2gmzuw/7xbpnIPTb+jcGupLlyNh15vfclQ0z9TxzC cSRyWSNT2VskmMURz42xKd/oVZnxVuZ0Kh4hORUFdtP6PxTVwc6MYTQw/Z9C9DoVQLNZs2JR 0i6TdWhADExSsg9w9gUY0ZyA9mtkwrD0DC0D7IXjbCLGZ80/rrA33frPcZ9ymvJ27c6j1U8X sRPMmimirZw9gjNCY/FiUGZl6GudagE3S7N8n2DzWuUs01CXg5wS6PIVm0caEvOtdn1+13OQ ru0Bbg6LgdMxtSOJ6VWZtHzg1hLS+/vNdfAb26rmmq9BAyEyq2DYYb2YWkQ0zjRBU4ZmA0N+ XaGKBM+Bju/rGLEETJuFFPubF30/OVirn6wU1U6wBuQb0J70rq44h4ViuKYS/MUxr8EvSggp yhuEFal2NLWD9qBpxZ/c6pAYdM95lZH1XvEtwNjJJCgKLpihlEGfwR2uUPhyQl3B51dncQ0q H4m1hd+Jb6X3V9beT6UwYr8N7PYJ2Xq+RCgca/W2lXQ0NaM/acP7ew1pUj/sg+0CkYv62tr3 tZO03SC/pjFDwUdXIrpUkYw8xh7p63abTch6IPU031sMLC0siLc1N4zA+sl0Bmgcs9FP6OAD gDyFdcWB9OyJ+wqnlipbw4EPPxJ+aEvOMOmav2G17K1M+Z7hj6miGtK4Y9n3kKL7Sp8RfbE3 5AezP6AxAuISy/8jEu9ssDwgY1IeCkeHnCjxij8GI5ReqpycJ4XBmi0Js23wsxyh5rsW35D6 F6uHEgG2c+seRqKdVPxxwxQ1UINoXymgyS01TJ0kyt65paYiWbK3OKoPE4DJWhjQHZky1HhP d7wx9sTWUztawkyiDOk41z7zu5VvvdRNW7WFG5Mdi6+FGpiV6qsqvLWaspG7tUzuiVSUfigS UidTq/hrhAa1SL6AmYYwyo0IWL58q7llgB33TrOZE14q2DULJoorf++zNnVRPoKmyEDWDE9k j7PQF61I9iu+9yQ0ZbFqOG3EWy7BdVIaSe+64SGuWOg4HFyRwWllqW6ntThVxMx0Sr6y8VCT SLMvQrxaYnt1r2nPKRgZEY7TETk5Z9CE5pl2pA1mIlW3HEbgpuP+n9Sql3IaYABhbnzOUgzE CYMx8/J7QPl3kx6M3/PwJj2A3yZy80nfNK6Z2IKxgo36N1MA6qPqrkYjW1yuFXr5RnJb61Fl ywGgeAr9GZcg+wNv181yT6BB7kJAURCFSn8zlKQ6NSvsKhcZGCuaKW9kk1kkrhNFZmkpQdRE Db8c5YmRmpr69lndUnLyDv1453lf9/ZaZQSsAeVmlHOlbodLpV5jfcMiSd9XAC19XQ41+42i wBv1pCmrcCGLWtq5qewHh9fMHX8ecoS/jjniasWkNyR2sijGZBoGzNDW5WNL7rgGTYbs7L8P AaLESEggm+cHavDEASf7kZ/sn+JFIqkdjmWKHQf0dR+VUyFPkUM5WJcFD4+n5M/CkWr3Jm4K hY/v2hNoAWg7EAQmYcKf1HlX2zSpRmlcGIxQZmbd19N6x1aolzSKYqY5/5yGCdR+tugqhaMI yqVfVctbylBV0qaClTkJrTr68PH9r3SPdCFd66UPa6D/N5gAu+PwYOz34Bm+TeVK8jJOWNtW vQ/00wFRnt5HsXFhx0FTDERnC/WKcvHtFG74CI9/aXduLz7HRni44eCEe4YONFm/VatiqeGN vKMrD5+LSdE25gMw37R1bVZ20QdwXILFXHlAfELsijDS7jVk6lcAksAai99A8BP6ro1wghHP cOzZsrd7rdjlbZ1Dl5EUQekgcS1fYkRJHn7MlrbBUGNPbDAJDvRwsixb7nuAbFXiexVsVW3t 1P5WwfYBA/bwmO1TB36A98ZlCafLQBTs4G7cw9wBC7kVt2zYxmyNplshj0zwKEornnNKGgXP CM6dh9d6LqK4kY6yr1zFndA4XxsMeSf03/Dvq+BcspQ6qMtWXg8nvkS+HkgzrpJ8CxIDOd4n ifftJ8LwRnuk+WCzCZmTAsbrz9KgIyRukAxca7d95RGRTPF5EdUtzTWUkxW4YA9TIaz6MUyg pDVman+KSlP6YfR9MoYXI3PLd6fdWEmKVzvESLVCw0MSXiqM3vejgpTiqL3lDXdo54kp5zrg JdLRKVcUQl/LckhUhE/QMcPdaVJCysjla+HgcUI43unsRSXQ99V65nDX/TUGv7vLTeFkZFOY AYOyr7ja4FPJsv8wUMoOTwY1MzaXlHdW9xAuHgrVTUP+BAXrGNzG18NjlribhKx7XQTE/+tg xNwjRFxNO0p/TGq+Fw3I1vWuAM6l1Q3ktj+xzXNYHj2Nqj6DuQ0Q2Lk8kM2NJ38WQN8awa/y FdlODnzTLVUl7J8dGpvhVyUqd5VFPVbV6EBfA4IyKTdeaAzyVoF4HbCpwcP9a7fBJBljgdvb ZO8sycKxVd4dNBsbe/RPPYblwIBwPvW4mnzjqZqnEddJl5RojrIKWhR4BBObeFgfHfNnKQkq g2ax2kSIS5VDaNs/Kgsrxtic6eB13yyieYbcxLtbuDHfarL5GHNyJzaHF9vhhFTzCwntfA10 N99IRDMDxl9kOKdT0ZQZ5ueJQwHPZUJqnSBInrR6a2IyJZxde1RD8jQRPSV/OYRi0OgR0MyG pgUq98GBt+q2V3ZKsHuKPgEzw8s7ULlPgfNAPNMcRON2DAJxqP3hIdwxpVYLyoBDH9VNDXuo K7Qoh42jfGDWtYvf3pcWZELfn47Q8y1nSdFsm8IVWPmlLtBlE7Zt3mn+mzZF1yeJ5J7ae2RZ A9wBd3+4jg5/6WsyBbW/pjYO2DmJIFit9vIuqsRo5eKDe8RTKEo6h2N3dAAGzrxCyiUTY3QR dC4cYQnYN3qB2zvV1W+j2lwVMLtJJO2KbDOhwj0RIFSuY3d3TY5NMb7GCtNfnU47+wF+q94Y hUOJpQhZhu9/T8EDPTqeluD34CcfTO1Lj9HU/RUzeO7fqFaiS02YbqzzHImCIoxz+y27VIlT pYXiBrT3rCmO5kYVjL8UC84GU2HtW8imm5tO/xniP842w/NuEIAPiqjcfwwLnRDu8AgCFiSJ 3RvF2d+QEWTx9mmgEbkz/UZ+C1TmMxR2OtOvS3lv5PRVzmrXbSitZTftydzJchjuaB6NpbvZ 9eXrJ6L1CKKV4He60fWNUzyX+ofgNVbJzhUBeVFiX1wc9JTopJPsAI4HoI3I7gFYEHNjrW6M Hx8CioD0SISV4WBxSEPxOCm1OmD/v91WIwlNA0YvZ5ChNoETiMwZTkR9vfLv2r+nWiZTGMGO 0EWsRQK4xgPxNYYQw== IronPort-Data: A9a23:vq1rXq/G1Vhsvdvqg6YiDrUDJnqTJUtcMsCJ2f8bNWPcYEJGY0x3y 2caDWvVa66CMGPxKo12O4y3/U9Q7JbXnINqQQFpqHpEQiMRo6IpJ/zJdxaqZ3v6wu7rFR88s Z1GMrEsCOhuExcwcz/0auCJQUFUjP3OHPymYAL9EngZbRd+Tys8gg5Ulec8g4p56fC0GArlV ena+qUzA3f7nWcqWo4ow/jb8k4346yi4GhwUmEWPJingneOzxH5M7pEfcldH1OgKqFIE+izQ fr0zb3R1gs1KD9wYj8Nuu+TnnwiGtY+DyDW4pZlc/TKbix5m8AH+v1T2Mzwxqtgo27hc9hZk L2hvHErIOsjFvWkdO81C3G0H8ziVEHvFXCuzXWX6KSuI0P6n3TE09F1MBAZJYMi++tUBm1jz KcRBTQQYUXW7w626OrTpuhEg9R6atHsOJIDt3pgyzDAEPtgRorMK0nIzYYCjXFp3pwIRq6YP JZxhTlHNHwsZzV0AGxPXchih+r9tEeqazpcuU6Yrqox4nHOwUp2yre4Od7cfJqRTsVQn1qEj mjB4mLyDwpcMYCPjz2f/RpAg8eWx3uqBN5MSuLQGvhCnnK+3Ug0UgwvRHCEgNuzpkW0Yo9kA hlBksYphfNvqBLwEYGVsweDiHWNuxpZX9tLD8Uh+QSVw+zV5RyYDy4KVFZ8hMcOsd9vAyQt0 k6VktjpAz12rbDTTmiSnluJkd+sETgxNlNTSjcKdwoIwN7vjbhohz/LU9k2RcZZkebJMT33x jmLqg03iLMSkdMH2s2HEbbv327ESn/hElZd2+nHYl9J+D+Vc6aLXeSVBbXz6OYZap6eSkids XMEncmH8e1ICouC/MBsfAnvNO/yjxpmGGSC6bKKI3XH32/zk5JEVdoAiAyS3G8zbq45lcbBO Sc/Qz956p5JJ2eNZqRqeY+3AMlC5fG/SI20D6uNNocRPMkZmOq7EMdGOR74M4fFzBhErE3DE czznTuEUytHUvoPIMSeG7hCidfHORzSNUuIGMyllEv/uVZvTHOeTrMIeEOJavEl4aiEpgTM7 tsXOtORwA03bQENSnS/zGLnFnhTdSJTLcmv+6Rqmhure1sO9JcJUq6JmetJlk0Mt/g9q9okC VnkCxEJmASk1SeaQehIA1g6AI7SsV9EhSpTFUQR0ZyAghDPuK7+vf9NRIh9ZrQ96u1owNh9S vRPKY3KAe1CRn6Ds34RZIX05t4qPhm6pxO8Dwz8ahgGfrlkW1Po/P3gdVDR7yUgNHe8mvY/h LyC7TnlZ6Q/aT5sNvuLV8L3/WiN5SAcvMlQQ3r3JsJifRSw0YpydA30oPwFA+ANDhThxzHB6 RunWykKgeyV+4IercfCgKuFi6yLEOJOOFVQMEeGzLSxNAjcpnGCx60ZWsm2XDntbkHG04T8W vd0ltbSaOYmmnROuKpCS4db97o0vYbTluUL3zZaE2XuRHX1LLFZe12t/9RF749JzZ9n4TqGY FqFoIRmCO/YKfHeMQAjISQ+Zb6+ztASoD7Z6Mo1LGjc5CNa+LmmU11YDyKTiR5yfadED4c4/ dgP4MImyRSzqh4PAOa0iip582etLHtZd44FspocIpHgiyt161VkTKHfNBTL48C0W40RCnUpH z6av7qdprJ+wkGZTWE/O0KQ1sVghLMPmitw8nk8G3qzlOHouNoLzTxK0DFuTg1q3hRNiO1yH W5wNnxKH6aF/hY2pc0aA0WXRh5LXg3E9mPPyVImyXXSf3escmmcPV8sGP2s+XoB+Dl2ZQlr/ 7C/yUfkXw31fcr34DAAZE59p9HnTv1z7gfniun+O+ikRr4UOSHEhI2qbko28yrXO9s73hD7l LM77dROZr3eHg9Oha8CUq2x96kaETKAL0x8GcBRxrsDRzzgSWvjyAq1Ch6Df+1WLKb36m6+M ctlI/xPWzmY1CqjqjM6B7YGE4RrncwGtcYzRbf2GVEo67evjCJlkJb1xBjMgGUGR9ZPk8FkD qjzczmEMHKbhFoKumvrgfRHBFGFYog/VFWh5Ny2zeQHLIJckedOdUpp7KC4kU/IOyRa/jWVn jj5WYno88JYx79BpbDcSpd4O13sKPfYdvi5zwSogtEfMfLNKZjvsi0WmHnGPiNXH6QbAcQqm ZuzstfYgVvOjIgyd2WIiquQNrJo4P+qV7F9KfPHL3h9nAqDVvTz4hAFxXuKFJxRnP5Z5eilX wGdavbsReUKWtxY+mJZWxJeHzkZFa7zSKXq/gG5kNihFTkf1lbhAO681Hq0c1xeSDAEC6f+B iDwpfyqwNJS96ZIJR0cAsBZE41KG0DiVYQmZu/Om2GhVEfwuWy7u5zmiRYEwhPIACPdEM/Fv LT0diKnfxG24Kz13NVVtrJphSIuDVF/vPIRe3wM8NsnmhG4C287dd4mC6skMa0NsCLO18DfX grvPVsSUXC3GXwOdBjn+93sUzuOHuFEaJ+zOjUt+FjScCusQp+JBLx67Cp7/nNqYX3Zwfq6L c0FsGjFVvRrLkqFmc5IjhB6vQtm+h8e7ncYoAbllMjjHxsVAbMLzWFsWg1XWkQr1unTwV7TK zFdqX9sGSmGpYzZSK6MuEK53DkTuy7pxDgzKyLT0JDYoYrzICho1qjkI++qulEcRJ1iGVPNL E8bg0OV5GSHxnEYua0oos8kx6huBppn2yR8wLDLHWUvok171ojr0w7uU8bCoAHONTOzy2/gq wQ= IronPort-HdrOrdr: A9a23:2gxonqGgk3UMYqDkpLqE1ceALOsnbusQ8zAXPiFKOH9om6mj/f xG88506faZslsssRIb+exoWpPgfZq0z/ccirX5Vo3MYOCJggeVBbAnxbSn6TztES/z+4dmpM VdWpk7Lsb/SXxzjcOS2njdLz/M+qjjzJyV X-Talos-CUID: 9a23:yqipImxjXQHUxJg1d+/9BgUEPtsZI1LFwk3RLmy/BUtQE7S4TUKfrfY= X-Talos-MUID: 9a23:nAkF9Ah89FyjRtTRnSfhccMpKuhF+ICJUlE0gL45muuOGSd6AmqmpWHi X-IronPort-Anti-Spam-Filtered: true X-IronPort-AV: E=Sophos;i="6.16,297,1744063200"; d="scan'208,217";a="120944667" X-MGA-submission: =?us-ascii?q?MDEJrsQGywEo7GgfV4Mjh04DlO208e6G9qFHuX?= =?us-ascii?q?mmzYOrNWEEffkPDfdSW+4TwtEnxpyCACR6Ai1VuI8Ytli6PZQ/UMuP1N?= =?us-ascii?q?k7Y5WVS/M/GEGOfv1zFCYfNFJjlVOU/5pmL9c77vamg9239iRsl5y7W6?= =?us-ascii?q?ZyulEOhILDecs2pbrUrujI+w=3D=3D?= Received: from mx1.polytechnique.org ([129.104.30.34]) by mail3-smtp-sop.national.inria.fr with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 08 Jul 2025 14:45:40 +0200 Received: from TM.local (unknown [82.66.240.41]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ssl.polytechnique.org (Postfix) with ESMTPSA id 6B59D564E70; Tue, 8 Jul 2025 14:45:38 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=polytechnique.org; s=svoboda; t=1751978738; bh=eZ4ag0Dh7yN9DLS4w6VmpATithEBbgfaovMHr+U2lO0=; h=From:To:Subject:Date:Message-ID; b=WNsKfbZxspbJKq72NwpHeHAY4RR7cI4LGLq8eFOGcNKLVEd45g5ikHVdSgf74WdhG 6ttAzxYNgybweXUjPaEQh+k3yLmniQdgsV3mxJ62hQHWf0x58Or6SBUmAJzHqPkLU7 oJ2LKocKB3JwTybZVTsolIf3Cn6chsJEoSGfEe+U= From: Alan Schmitt To: "lwn" , caml-list@inria.fr Date: Tue, 08 Jul 2025 14:45:36 +0200 Message-ID: MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="=-=-=" X-AV-Checked: ClamAV using ClamSMTP at svoboda.polytechnique.org (Tue Jul 8 14:45:39 2025 +0200 (CEST)) X-Spam-Flag: Unsure, tests=bogofilter, spamicity=0.499962, queueID=AAD0F564E71 X-Org-Mail: alan.schmitt.1995@polytechnique.org Subject: [Caml-list] Attn: Development Editor, Latest OCaml Weekly News Reply-To: Alan Schmitt X-Loop: caml-list@inria.fr X-Sequence: 19350 Errors-To: caml-list-owner@inria.fr Precedence: list Precedence: bulk Sender: caml-list-request@inria.fr X-no-archive: yes List-Id: List-Help: List-Subscribe: List-Unsubscribe: List-Post: List-Owner: List-Archive: Archived-At: --=-=-= Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Hello Here is the latest OCaml Weekly News, for the week of July 01 to 08, 2025. Table of Contents =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80 OCaml security team Dockerfile for building an OPAM application, with dune/apt/opam caching opam 2.4.0~rc1 Js_of_ocaml 6.1.0 / Wasm_of_ocaml Announcing Raven: Scientific Computing for OCaml (Alpha Release) Slipshow! Other OCaml News Old CWN OCaml security team =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90 Archive: Hannes Mehnert announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 Dear everyone, We are starting an effort to push security into OCaml. This is based on discussions in the OCaml Software Foundation with industry partners. The main goal is to have best practises similar to those of other programming language ecosystems. Reporting security issues =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C This entails a point of contact for the security team - which deals with communication between the person who found a security-relevant problem in OCaml software (named "reporter"), who can then contact us - the security team - instead of using a public bug tracker, and the upstream OCaml developer(s). We, the security team, will establish the three-way communication, and since we have a documented security disclosure process (which will be published soon), we will guide everyone through the process and ensure that timelines are met, CVE numbers are assigned, =E2=80=A6 Team composition =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C The OCaml security team currently consists of individual security experts and individuals representing company sponsors of the OCaml Software Foundation. Individual members participating on a personal capacity may be compensated for their time from the OCaml Software Foundation. The team currently consists of 7 members =E2=80=A2 Hannes Mehnert - individual, chair =E2=80=A2 Mindy - individual =E2=80=A2 Joe - individual =E2=80=A2 Edwin T=C3=B6r=C3=B6k - individ= ual =E2=80=A2 Nicol=C3=A1s Ojeda B=C3=A4r - LexiFi =E2=80=A2 Louis Roch=C3=A9 - ahrefs =E2=80=A2 Maxim Grankin - Bloomberg We're in the process to formalise the responsibilities of the team, our proposed disclosure process, and how to join & leave the team. Funding for security actions =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C= =E2=95=8C=E2=95=8C=E2=95=8C In complement to the security disclosure process, we will accept funding requests for projects that make OCaml more secure (including guidelines how to develop OCaml in a secure way/what are common pitfalls; static analysis; dissemination tools; =E2=80=A6). The OCaml Sof= tware Foundation will provide funding for these security actions. After this summer we will discuss this in more depth with the community. Next steps =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C We will setup a website (similar to has) soon, and provide an email address for contacting us - security At ocamlDoT org is forwarding to our team. We plan to setup a mailing list for security announcements. But more on that at a later point, this brief post is mainly about the fact that this team starts to exist now, and is working on improving the security story of OCaml. If you have any questions for now, please feel free to discuss them in this announcement. Please be aware that it is vacation time soon, so we may not be very responsive. Dockerfile for building an OPAM application, with dune/apt/opam caching =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90= =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90= =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 Archive: Edwin T=C3=B6r=C3=B6k announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 I was writing some dockerfiles for building some OCaml applications recently, and realized that it is possible to write a Dockerfile that can build almost arbitrary Opam applications without even having to hardcode the package name. I used `ocaml-dockerfile' to generate it (only the OS and OCaml version is hardcoded), so I thought I'd share it, it should work with both Podman and Docker: =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 # syntax=3Ddocker/dockerfile:1 =E2=94=82=20 =E2=94=82 FROM ocaml/opam:debian-12-ocaml-4.14 =E2=94=82=20 =E2=94=82 # Update the opam repository (otherwise lockfile may contain a = version we don't know about) =E2=94=82 RUN git -C /home/opam/opam-repository pull origin master && opa= m-2.3 update =E2=94=82=20 =E2=94=82 # Enable the dune cache in copy mode =E2=94=82 # (hardlink mode would fail with EXDEV if cache is on separate = disk/partition) =E2=94=82 # This caches the build of opam dependencies using dune and the= main application =E2=94=82 ENV DUNE_CACHE=3D"enabled" DUNE_CACHE_STORAGE_MODE=3D"copy" =E2=94=82=20 =E2=94=82 # `apt` in containers is usually configured to always clean up = after operations =E2=94=82 # When we have a cache mount for the APT package cache we don't= want to clean up, since the cache would be ineffective. =E2=94=82 # Also change OPAM to use the 0install solver, this is faster, = especially when a lockfile is present. =E2=94=82 RUN sudo rm -f /etc/apt/apt.conf.d/docker-clean && \ =E2=94=82 opam-2.3 option solver=3Dbuiltin-0install =E2=94=82=20 =E2=94=82 # Use a workdir outside of $HOME, to avoid having .opam as a su= bdir of the build =E2=94=82 WORKDIR /app =E2=94=82 # Copy dependency definitions first. See https://docs.docker.co= m/build/cache/optimize/#order-your-layers =E2=94=82 COPY [ "*.opam", "*.opam.locked", "." ] =E2=94=82=20 =E2=94=82 # Install and cache system packages required by the build =E2=94=82 # The cache is locked, so multiple container builds will wait h= ere =E2=94=82 # (apt would have a lock but it is stored outside of the cache = dir, so we cannot rely on it to prevent concurrent accesses) =E2=94=82 # A cache id is used, so that different distros would have diff= erent cache folders =E2=94=82 RUN --mount=3Dtype=3Dcache,id=3D/var/cache/apt#debian-12;amd64,= target=3D/var/cache/apt,sharing=3Dlocked \ =E2=94=82 --mount=3Dtype=3Dcache,id=3D/var/lib/apt#debian-12;amd64,ta= rget=3D/var/lib/apt,sharing=3Dlocked \ =E2=94=82 sudo apt-get update -y && \ =E2=94=82 opam-2.3 install --locked --with-test . --depext-only =E2=94=82=20 =E2=94=82 # Download and cache opam package dependencies =E2=94=82 # The cache is locked, so multiple container builds will wait h= ere =E2=94=82 # To minimize the time the lock is held the actual package inst= allations are done as a separate step =E2=94=82 RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.opam/download-c= ache,sharing=3Dlocked,uid=3D1000,gid=3D1000 \ =E2=94=82 opam-2.3 install --locked --with-test . --download-only =E2=94=82=20 =E2=94=82 # Install (cached) downloaded opam dependencies =E2=94=82 # The download cache is mounted readonly and shared, multiple c= ontainer builds can proceed in parallel. =E2=94=82 # The dune cache is mounted RW. =E2=94=82 # Multiple concurrent builds will use the same cache, but dune = must already be able to cope with this =E2=94=82 # A cache id is used, so that different distros would have diff= erent cache folders (it is unlikely that dune caches would be sharable acro= ss distros) =E2=94=82 RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.opam/download-c= ache,readonly,sharing=3Dshared,uid=3D1000,gid=3D1000 \ =E2=94=82 --mount=3Dtype=3Dcache,target=3D/home/opam/.cache/dune,shar= ing=3Dshared,uid=3D1000,gid=3D1000 \ =E2=94=82 opam-2.3 install --locked --with-test . --deps-only =E2=94=82=20 =E2=94=82 # Copy actual application source code =E2=94=82 COPY [ ".", "." ] =E2=94=82=20 =E2=94=82 # Build and install application code, using dune cache. =E2=94=82 RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.cache/dune,shar= ing=3Dshared,uid=3D1000,gid=3D1000 \ =E2=94=82 opam-2.3 install --locked --with-test . =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 It avoids some common pitfalls: =E2=80=A2 disable cleanup of APT downloaded packages (otherwise caching is ineffective) =E2=80=A2 enables the opam download cache too which is stored in a non-standard location (`~/.opam/download-cache' instead of `~/.cache/opam') =E2=80=A2 avoids EXDEV from dune cached builds during opam dependency installations by enabling dune cache copy mode =E2=80=A2 uses opam-2.3 explicitly, since the default opam in the Dockerh= ub images is quite old (luckily opam-2.3 is already there, just needs to be invoked explicitly) =E2=80=A2 sets the default solver to 0install. This seems to be needed to speed up `opam install' even when `--locked' is used and no dependency resolution is needed. Otherwise it was spending 4s checking the solver request. =E2=80=A2 sets appropriate uid in cache mounts to avoid permission issues Caveats: =E2=80=A2 if you use git submodules then you have to add `.git' to your `.dockerignore'. Otherwise opam pinning will fail, since `.git' is a file referencing a git dir in a parent dir that is not mounted inside the docker build environment =E2=80=A2 maybe =E2=80=93dev should be used if you intend to use the cont= ainer for developing the opam application. Although in that case you might also want to preinstall some useful tools like `lsp', and `ocamlformat'. Eventually I hope this can be simplified using Dune's new package management feature. opam 2.4.0~rc1 =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 Archive: Kate announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 Hi everyone, We are happy to announce the first release candidate of opam 2.4.0. This is hopefully the first and last release candidate for opam 2.4, so we invite users to test it to spot previously unnoticed bugs as we soon head to the stable release. Changes =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C =E2=80=A2 :woman_scientist: Fix a regression in `opam switch create ' not working when all compilers of that version are flagged with `avoid-version'. This would have prevented users to use commands such as `opam switch create 5.4.0~alpha1' ([#6563]). =E2=80=A2 :high_speed_train: Improve performance of `opam update' for use= rs of local repositories that happen to be git repositories (for example if you've ever used `opam repository add --kind local' or similar). In particular the new OCaml implementation of patch, does not scan those VCS directories anymore, which made opam use a lot of RAM unnecessarily ([#6560]). :open_book: You can read our [blog post] to read about the other more minor changes, and for even more details you can take a look at the [release note] or the [changelog]. [#6563] [#6560] [blog post] [release note] [changelog] Try it! =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C The upgrade instructions are unchanged: For Unix systems =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 bash -c "sh <(curl -fsSL https://opam.ocaml.org/install.sh) --v= ersion 2.4.0~rc1" =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 or from PowerShell for Windows systems =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 Invoke-Expression "& { $(Invoke-RestMethod https://opam.ocaml.o= rg/install.ps1) } -Version 2.4.0~rc1" =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 Please report any issues to the [bug-tracker]. Happy hacking, <> <> The opam team <> <> :camel: [bug-tracker] Js_of_ocaml 6.1.0 / Wasm_of_ocaml =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90= =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 Archive: Hhugo announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 I=E2=80=99m pleased to announce the joint release of js_of_ocaml 6.1.0 and wasm_of_ocaml. Js_of_ocaml is a compiler from OCaml bytecode to JavaScript. It makes it possible to run pure OCaml programs in JavaScript environment like browsers and Node.js. [Wasm_of_ocaml] is a compiler from OCaml bytecode to WebAssembly. It is highly compatible with Js_of_ocaml, so you can compile your programs with wasm_of_ocaml instead of js_of_ocaml and experience overall better performance. Most significant changes: =E2=80=A2 A lot of effort was spent on optimizing compilation speed and compile-time memory usage. we're seeing up to 4x compilation speed improvement in some cases. See the [Changelog] for other changes. [Wasm_of_ocaml] [Changelog] Announcing Raven: Scientific Computing for OCaml (Alpha Release) =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90= =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95= =90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90= =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 Archive: Thibaut Mattio announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 I'm excited to announce the alpha release of [Raven], a modern scientific computing ecosystem for OCaml. [Raven] What is Raven? =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C Raven is a collection of libraries and tools for numerical computing and machine learning, including: =E2=80=A2 *Nx:* Multi-dimensional arrays with NumPy-like operations and pluggable backends (now pure OCaml, C FFI, Metal, next CUDA, WebGPU, etc.) - our equivalent of NumPy =E2=80=A2 *Rune:* Automatic differentiation and device placement, building toward JIT compilation - our equivalent of Jax =E2=80=A2 *Kaun:* Deep learning framework inspired by [Flax]/[PyTorch], b= uilt on Rune =E2=80=A2 *Sowilo:* Computer vision library with differentiable operation= s, build on Rune =E2=80=A2 *Hugin:* Plotting library for data visualization - our equivale= nt of matplotlib =E2=80=A2 *Quill:* Markdown-first interactive notebooks - very different = from Jupyter, but our answer to interactive notebooks The ecosystem is designed to work together seamlessly, with Nx as the foundation, Rune providing differentiable computation, and domain-specific libraries building on top. [Flax] [PyTorch] Getting Started =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C Install Raven via opam: =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 opam install raven =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 Here's a quick example showcasing automatic differentiation with Rune: =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 open Rune =E2=94=82=20 =E2=94=82 (* Define a simple neural network layer *) =E2=94=82 let layer ~w ~b x =3D add (matmul x w) b =E2=94=82=20 =E2=94=82 (* Compute mean squared error loss *) =E2=94=82 let mse_loss ~w ~b x y =3D =E2=94=82 let pred =3D layer ~w ~b x in =E2=94=82 let diff =3D sub pred y in =E2=94=82 mean (mul diff diff) =E2=94=82=20 =E2=94=82 let result =3D =E2=94=82 (* Choose device - Rune.ocaml, Rune.c, Rune.metal *) =E2=94=82 let dev =3D Rune.c in =E2=94=82=20 =E2=94=82 (* Initialize parameters on the device *) =E2=94=82 let w =3D randn dev float32 [| 3; 2 |] in =E2=94=82 let b =3D zeros dev float32 [| 2 |] in =E2=94=82 let x =3D randn dev float32 [| 10; 3 |] in =E2=94=82 let y =3D randn dev float32 [| 10; 2 |] in =E2=94=82=20 =E2=94=82 (* Compute loss and gradients *) =E2=94=82 let loss, grad_w =3D value_and_grad (fun w -> mse_loss ~w ~b = x y) w in =E2=94=82 Printf.printf "Loss: %g\n" (unsafe_get [] loss); =E2=94=82 grad_w =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 val result : (float, Rune.float32_elt, [ `c ]) Rune.t =3D =E2=94=82 [[-1.74967, 0.863766], =E2=94=82 [-0.140407, -0.269364], =E2=94=82 [0.593187, 0.0197736]] =E2=94=82 Loss: 2.13033 =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 For more examples and detailed documentation, visit [raven-ml.dev]. [raven-ml.dev] Why Raven? =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C Today's machine learning ecosystem is converging on frameworks built atop ML compilers: high-level Python APIs that build computation graphs, then JIT compile them for performance. In parallel, [JAX] and [Flax] have gained popularity with their functional APIs. For instance, JAX uses function transformations (grad, jit, vmap) to implements its core features. This is a landscape where OCaml has natural advantages. OCaml excels at building compilers, which includes ML compilers, and as a functional language, it's a more natural fit for functional ML APIs than Python with JAX. Given these technical advantages, Python's dominance comes down to developer experience: the massive ecosystem and excellent prototyping ergonomics. We believe that with the right tooling, OCaml can match Python's productivity for prototyping and exploratory work. And the ecosystem gap doesn't have any fundamental challenge: we "just" need to write a lot of code. If Raven succeeds, we believe it will offer a much more compelling alternative to Python: a language that enables rapid prototyping while eliminating the gap between exploration and production. You'll move from local development to production without switching languages, without separate teams, without maintaining two stacks. [JAX] [Flax] Technical Highlights: Rune's autodiff engine =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C= =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C One interesting aspect of Raven is Rune's implementation of automatic differentiation using OCaml's effects system. As far as we know, this is the first production-scale autodiff engine built on effects, drawing on research by [Jesse Sigal] and earlier work by [KC Sivaramakrishnan]. The architecture follows a modular design: =E2=80=A2 Pluggable backends in Nx allow implementation for different har= dware =E2=80=A2 Rune implements an Nx backend that raises effects for all operations. =E2=80=A2 These effects are either caught by an effect handler (e.g. grad, jit), or, if unhandled, executed eagerly =E2=80=A2 This allows for composable effects handlers (e.g. grad (grad f)= , jit (grad f), etc.) [Jesse Sigal] [KC Sivaramakrishnan] Technical Highlights: Quill Notebooks =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C= =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C Quill reimagines interactive notebooks with a [Typora]like experience. Markdown renders live as you write, switching to raw markdown when you focus on a section for editing. This creates a natural writing experience where code blocks integrate naturally into your document. The result is a distraction-free notebook environmnet, that prioritizes focused writing, while still providing full editor features within code blocks (coming soon =E2=84=A2!). While still early, we're excited to see how the community reacts to Quill when it is stable enough for daily use - we really think it has the potential to offer a much better notebook experience for teaching, reading, and other workflows. [Typora] Current Status =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C This is an alpha release. APIs are stabilizing but may still change. Things will break - this is expected at this stage! If you encounter bugs, please open an issue on GitHub; community feedback is invaluable to get to a stable release. We're currently focused on: =E2=80=A2 Stabilizing core APIs for the 1.0 release =E2=80=A2 Writing documentation and user guides =E2=80=A2 Supporting early users adoption (FFT, linear algebra, CUDA back= end) Post-Alpha priorities include JIT compilation and stable Quill environment. Building a Community =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C One bet we're taking with Raven is that it will allow a scientific and ML community in OCaml to flourish. As of now, it's still largely a one-person project. While I'm committed to its development, we really need to see the development of a larger community for a project of this size to survive. If you're interested in the project, the best thing you can do is to engage=E2=80=94whether by opening issues, reaching out, or contributing. Alongside reaching a first stable release, building a community is Raven's main priority from now on, so any kind of contribution or engagement will be deeply appreciated. If there's anything I can do to make Raven more welcoming and approachable, let me know. I've always believed that the best way to grow OCaml adoption is to provide killer apps for specific use cases (just like Rails did for Ruby). Raven's not quite there yet in terms of advantages over Python, but it can get there, and if that's something you'd like to contribute to, please reach out\! Getting Involved =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C Here are immediate ways to contribute as we work toward a stable release: *For users:* =E2=80=A2 Try the libraries with your workflows and report issues =E2=80=A2 Share feedback on API design and usability =E2=80=A2 Help test on different platforms and configurations *For contributors:* =E2=80=A2 Optimize eager execution backends while we build JIT compilation =E2=80=A2 Add missing NumPy/Jax/Flax operations to Nx/Rune/Kaun =E2=80=A2 Contribute examples and documentation Don't hesitate to reach out if you'd like to be involved closely with the core development: =E2=80=A2 JIT compilation =E2=80=A2 Stabilize Quill (many bugs to fix!) =E2=80=A2 New libraries *Resources:* =E2=80=A2 GitHub: =E2=80=A2 Documentation: =E2=80=A2 Contact: thibaut.mattio@gmail.com Acknowledgments =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C I'd like to thank our early contributors and testers who have helped shape Raven: =E2=80=A2 @axrwl =E2=80=A2 @gabyfle =E2=80=A2 @hesterjeng =E2=80=A2 @ghennequin =E2=80=A2 @kayceesrk =E2=80=A2 @blueavee Special thanks to our GitHub sponsors for their support: =E2=80=A2 @daemonfire300 =E2=80=A2 @gabyfle =E2=80=A2 @sabine Your feedback, contributions and support have been invaluable in getting Raven to this alpha release - thank you\! Supporting Raven's Development =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C= =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C As Raven grows, I'm looking for sustainable ways to continue dedicating time to the project. If you're an industrial user interested in using Raven for your machine learning or scientific computing needs, I'd love to talk about how we can work together. For individuals who want to support the project, I have a [GitHub Sponsors page]. Any contribution, no matter the size, is deeply appreciated and helps ensure Raven's continued development. [GitHub Sponsors page] Slipshow! =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90 Archive: Continuing this thread, Paul-Elliot announced =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80= =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80 Let's continue with a new release in this Slipshow thread. It's my _ginormous pleasure_ to announce the [opam release] of: [opam release] Slipshow v0.3.0: The return of the subslips =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C= =E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2= =95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95=8C=E2=95= =8C=E2=95=8C As you can see, a crucial improvement is that releases are now named. Semver is boring, let's add a bit of fun in a serious world. The release name subtly suggests that subslips are back. Yes, subslips are back! Slipshow allows your presentation to be visually organized. Subslips are slips that are inside a slip. You can then "enter" it, go through it, and exit it to go back to the original slip. But a gif is worth a ginormous word, so here it is: [A demo of slipshow entering subslips] To obtain a similar effect, using the new release, you can simply do: =E2=94=8C=E2=94=80=E2=94=80=E2=94=80=E2=94=80 =E2=94=82 We will discuss three topics: =E2=94=82=20 =E2=94=82 {style=3D"display:flex" children:slip} =E2=94=82 ---- =E2=94=82 # Topic 1 =E2=94=82=20 =E2=94=82 Content of the first topic =E2=94=82=20 =E2=94=82 --- =E2=94=82 # Topic 2 =E2=94=82=20 =E2=94=82 Content of the second topic =E2=94=82=20 =E2=94=82 --- =E2=94=82 # Topic 3 =E2=94=82=20 =E2=94=82 Content of the third topic =E2=94=94=E2=94=80=E2=94=80=E2=94=80=E2=94=80 If you want to separate the source in multiple file, it's easy: `{include src=3D"file/to/include.md"}'. And did you notice that `---' now act as group separators? At this point, let me just output the changelog (highligting some of them): [A demo of slipshow entering subslips] =E2=97=8A Compiler =E2=80=A2 *Fix file watching issues* by vendoring a (modified) irmin-watc= her, and watching all files the presentation depends on (images, themes, =E2=80=A6) (#113) =E2=80=A2 Adds a favicon to the presentation file (*Slipshow now has a lo= go!*) (#115) =E2=80=A2 Fix missing attributes on images (#117) =E2=80=A2 Fix missing mime type on images that made svg undisplayable (#1= 20) =E2=80=A2 Fix detection of math inside inline attributes (#124) =E2=80=A2 *Add `--dimension' to specify the dimension of the presentation (#131)* =E2=80=A2 Add less boring name for versions (#132) =E2=97=8A Language =E2=80=A2 *Add `{include src=3D"path/to/file.md"}' to include a file in a= nother (#114)* =E2=80=A2 Allow `pause' to have a target (#118) =E2=80=A2 Remove the need for `step' to execute actions (#118) =E2=80=A2 *Added support for subslips and slides (#118)* =E2=80=A2 Added pause blocks (#127) =E2=80=A2 *Use horizontal lines (`---') to group blocks (#129)* =E2=80=A2 Pass attributes to children with `children:' (#130) =E2=80=A2 Consistently remove the need for `-at-unpause' (#133) =E2=97=8A Engine =E2=80=A2 Simplify table of content by removing preview (#118) =E2=80=A2 *Fix wrong computation of location (#118, #119)* =E2=80=A2 *Improve zooming behaviour (everywhere) and performance (on chrome-based browsers=E2=80=A6) (#121)* =E2=80=A2 If someone has some expertise on how to improve performance on firefox, I'm interested! =E2=80=A2 Add PageUp and PageDown as navigation keys, adding support for pointers (#126) =E2=80=A2 Do not act when control is pressed (#126) =E2=80=A2 Fix wrong positioning on scaled slips (#128) =E2=97=8A Credits Thanks to the NLNet foundation for supporting this project! Other OCaml News =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2= =95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 >>From the ocaml.org blog =E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2= =94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94= =80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80=E2=94=80 Here are links from many OCaml blogs aggregated at [the ocaml.org blog]. =E2=80=A2 [Weeks 24-27] =E2=80=A2 [The week that was - 2025 w27] =E2=80=A2 [Improving Memory Profiler Visualisations for OCaml] =E2=80=A2 [The Hell of Tetra Master] [the ocaml.org blog] [Weeks 24-27] [The week that was - 2025 w27] [Improving Memory Profiler Visualisations for OCaml] [The Hell of Tetra Master] Old CWN =E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90=E2=95=90 If you happen to miss a CWN, you can [send me a message] and I'll mail it to you, or go take a look at [the archive] or the [RSS feed of the archives]. If you also wish to receive it every week by mail, you may subscribe to the [caml-list]. [Alan Schmitt] [send me a message] [the archive] [RSS feed of the archives] [caml-list] [Alan Schmitt] --=-=-= Content-Type: text/html; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable OCaml Weekly News

OCaml Weekly News

Previous Week<= /a> Up Next Week

Hello

Here is the latest OCaml Weekly News, for the week of July 01 to 08, 2025.

OCaml security team

Archive: https://discuss.ocaml.org/t/ann-ocaml-security-team/16902/1

Hannes Mehnert announced

Dear everyone,

We are starting an effort to push security into OCaml. This is based on dis= cussions in the OCaml Software Foundation with industry partners. The main = goal is to have best practises similar to those of other programming langua= ge ecosystems.

Reporting security issues

This entails a point of contact for the security team - which deals with co= mmunication between the person who found a security-relevant problem in OCa= ml software (named "reporter"), who can then contact us - the security team= - instead of using a public bug tracker, and the upstream OCaml developer(= s).

We, the security team, will establish the three-way communication, and sinc= e we have a documented security disclosure process (which will be published= soon), we will guide everyone through the process and ensure that timeline= s are met, CVE numbers are assigned, …

Team composition

The OCaml security team currently consists of individual security experts a= nd individuals representing company sponsors of the OCaml Software Foundati= on. Individual members participating on a personal capacity may be compensa= ted for their time from the OCaml Software Foundation.

The team currently consists of 7 members

We're in the process to formalise the responsibilities of the team, our pro= posed disclosure process, and how to join & leave the team.

Funding for security actions

In complement to the security disclosure process, we will accept funding re= quests for projects that make OCaml more secure (including guidelines how t= o develop OCaml in a secure way/what are common pitfalls; static analysis; = dissemination tools; …). The OCaml Software Foundation will provide = funding for these security actions. After this summer we will discuss this = in more depth with the community.

Next steps

We will setup a website (similar to https://www.haskell.org/security/ has) soon, and provide an emai= l address for contacting us - security At ocamlDoT org is forwarding to our= team. We plan to setup a mailing list for security announcements.

But more on that at a later point, this brief post is mainly about the fact= that this team starts to exist now, and is working on improving the securi= ty story of OCaml.

If you have any questions for now, please feel free to discuss them in this= announcement. Please be aware that it is vacation time soon, so we may not= be very responsive.

Dockerfile for building an OPAM application, with dune/apt/opa= m caching

Archive: https://discuss.ocaml.= org/t/dockerfile-for-building-an-opam-application-with-dune-apt-opam-cachin= g/16906/1

Edwin T=C3=B6r=C3=B6k announced

I was writing some dockerfiles for building some OCaml applications recentl= y, and realized that it is possible to write a Dockerfile that can build al= most arbitrary Opam applications without even having to hardcode the packag= e name.

I used ocaml-dockerfile to generate it (only the OS and OCaml = version is hardcoded), so I thought I'd share it, it should work with both = Podman and Docker: 

# syntax=3Ddocker/dockerfile:1

FROM ocaml/opam:debian-12-ocaml-4.14

# Update the opam repository (otherwise lockfile may contain a version we d=
on't know about)
RUN git -C /home/opam/opam-repository pull origin master && opam-2.=
3 update

# Enable the dune cache in copy mode
# (hardlink mode would fail with EXDEV if cache is on separate disk/partiti=
on)
# This caches the build of opam dependencies using dune and the main applic=
ation
ENV DUNE_CACHE=3D"enabled" DUNE_CACHE_STORAGE_MODE=3D"copy"

# `apt` in containers is usually configured to always clean up after operat=
ions
# When we have a cache mount for the APT package cache we don't want to cle=
an up, since the cache would be ineffective.
# Also change OPAM to use the 0install solver, this is faster, especially w=
hen a lockfile is present.
RUN sudo rm -f /etc/apt/apt.conf.d/docker-clean && \
  opam-2.3 option solver=3Dbuiltin-0install

# Use a workdir outside of $HOME, to avoid having .opam as a subdir of the =
build
WORKDIR /app
# Copy dependency definitions first. See https://docs.docker.com/build/cach=
e/optimize/#order-your-layers
COPY [ "*.opam", "*.opam.locked", "." ]

# Install and cache system packages required by the build
# The cache is locked, so multiple container builds will wait here
# (apt would have a lock but it is stored outside of the cache dir, so we c=
annot rely on it to prevent concurrent accesses)
# A cache id is used, so that different distros would have different cache =
folders
RUN --mount=3Dtype=3Dcache,id=3D/var/cache/apt#debian-12;amd64,target=3D/va=
r/cache/apt,sharing=3Dlocked \
    --mount=3Dtype=3Dcache,id=3D/var/lib/apt#debian-12;amd64,target=3D/var/=
lib/apt,sharing=3Dlocked \
    sudo apt-get update -y && \
    opam-2.3 install --locked --with-test . --depext-only

# Download and cache opam package dependencies
# The cache is locked, so multiple container builds will wait here
# To minimize the time the lock is held the actual package installations ar=
e done as a separate step
RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.opam/download-cache,sharing=
=3Dlocked,uid=3D1000,gid=3D1000 \
    opam-2.3 install --locked --with-test . --download-only

# Install (cached) downloaded opam dependencies
# The download cache is mounted readonly and shared, multiple container bui=
lds can proceed in parallel.
# The dune cache is mounted RW.
# Multiple concurrent builds will use the same cache, but dune must already=
 be able to cope with this
# A cache id is used, so that different distros would have different cache =
folders (it is unlikely that dune caches would be sharable across distros)
RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.opam/download-cache,readonl=
y,sharing=3Dshared,uid=3D1000,gid=3D1000 \
    --mount=3Dtype=3Dcache,target=3D/home/opam/.cache/dune,sharing=3Dshared=
,uid=3D1000,gid=3D1000 \
    opam-2.3 install --locked --with-test . --deps-only

# Copy actual application source code
COPY [ ".", "." ]

# Build and install application code, using dune cache.
RUN --mount=3Dtype=3Dcache,target=3D/home/opam/.cache/dune,sharing=3Dshared=
,uid=3D1000,gid=3D1000 \
    opam-2.3 install --locked --with-test .

It avoids some common pitfalls:

  • disable cleanup of APT downloaded packages (otherwise caching is ineffe= ctive)
  • enables the opam download cache too which is stored in a non-standard l= ocation (~/.opam/download-cache instead of ~/.cache/opam= )
  • avoids EXDEV from dune cached builds during opam dependency installatio= ns by enabling dune cache copy mode
  • uses opam-2.3 explicitly, since the default opam in the Dockerhub image= s is quite old (luckily opam-2.3 is already there, just needs to be invoked= explicitly)
  • sets the default solver to 0install. This seems to be needed to speed u= p opam install even when --locked is used and no = dependency resolution is needed. Otherwise it was spending 4s checking the = solver request.
  • sets appropriate uid in cache mounts to avoid permission issues

Caveats:

  • if you use git submodules then you have to add .git to you= r .dockerignore. Otherwise opam pinning will fail, since .git is a file referencing a git dir in a parent dir that is not mo= unted inside the docker build environment
  • maybe –dev should be used if you intend to use the container for= developing the opam application. Although in that case you might also want= to preinstall some useful tools like lsp, and ocamlform= at.

Eventually I hope this can be simplified using Dune's new package managemen= t feature.

opam 2.4.0~rc1

Archive: https://discuss.ocaml.org/t/ann-opam-2-4-0-rc1/16910/1

Kate announced

Hi everyone,

We are happy to announce the first release candidate of opam 2.4.0.

This is hopefully the first and last release candidate for opam 2.4, so we = invite users to test it to spot previously unnoticed bugs as we soon head t= o the stable release.

Changes

  • :woman_scientist: Fix a regression in opam switch create <vers= ion> not working when all compilers of that version are flagged w= ith avoid-version. This would have prevented users to use comm= ands such as opam switch create 5.4.0~alpha1 (#6563).
  • :high_speed_train: Improve performance of opam update for = users of local repositories that happen to be git repositories (for example= if you've ever used opam repository add --kind local or simil= ar). In particular the new OCaml implementation of patch, does not scan tho= se VCS directories anymore, which made opam use a lot of RAM unnecessarily = (#6560).

:open_book: You can read our blog post to read about the other more minor changes, and for= even more details you can take a look at the release note or the changelog.

Try it!

The upgrade instructions are unchanged:

For Unix systems 

bash -c "sh <(curl -fsSL https://opam.ocaml.org/install.sh) --version 2.=
4.0~rc1"

or from PowerShell for Windows systems 

Invoke-Expression "& { $(Invoke-RestMethod https://opam.ocaml.org/insta=
ll.ps1) } -Version 2.4.0~rc1"

Please report any issues to the bug-tracker.

Happy hacking, <> <> The opam team <> <> :camel:

Js_of_ocaml 6.1.0 / Wasm_of_ocaml

Archive: https://discuss.ocaml.org/t/ann-js-of-ocaml-6-1-0-wasm-of= -ocaml/16912/1

Hhugo announced

I=E2=80=99m pleased to announce the joint release of js_of_ocaml 6.1.0 and = wasm_of_ocaml.

Js_of_ocaml is a compiler from OCaml bytecode to JavaScript. It makes it po= ssible to run pure OCaml programs in JavaScript environment like browsers a= nd Node.js.

Wasm_of= _ocaml is a compiler from OCaml bytecode to WebAssembly. It is highly c= ompatible with Js_of_ocaml, so you can compile your programs with wasm_of_o= caml instead of js_of_ocaml and experience overall better performance.

Most significant changes:

  • A lot of effort was spent on optimizing compilation speed and compile-t= ime memory usage. we're seeing up to 4x compilation speed improvement in so= me cases.

See the Changelog for other changes.

Announcing Raven: Scientific Computing for OCaml (Alpha Releas= e)

Archive: https://discuss.ocaml.org/t/ann= ouncing-raven-scientific-computing-for-ocaml-alpha-release/16913/1

Thibaut Mattio announced

I'm excited to announce the alpha release of Raven, a modern scientific computing ecosystem for OCam= l.

What is Raven?

Raven is a collection of libraries and tools for numerical computing and ma= chine learning, including:

  • Nx: Multi-dimensional arrays with NumPy-like operations and plug= gable backends (now pure OCaml, C FFI, Metal, next CUDA, WebGPU, etc.) - ou= r equivalent of NumPy
  • Rune: Automatic differentiation and device placement, building t= oward JIT compilation - our equivalent of Jax
  • Kaun: Deep learning framework inspired by Flax/PyTorch,= built on Rune
  • Sowilo: Computer vision library with differentiable operations, = build on Rune
  • Hugin: Plotting library for data visualization - our equivalent = of matplotlib
  • Quill: Markdown-first interactive notebooks - very different fro= m Jupyter, but our answer to interactive notebooks

The ecosystem is designed to work together seamlessly, with Nx as the found= ation, Rune providing differentiable computation, and domain-specific libra= ries building on top.

Getting Started

Install Raven via opam: 

opam install raven

Here's a quick example showcasing automatic differentiation with Rune: 

open Rune

(* Define a simple neural network lay=
er *)
let layer ~w=
 ~b =
x =3D add (matmul x w) b

(* Compute mean squared error loss *)
let mse_loss ~w ~b x y =3D
  let pred =3D layer ~w ~b x in
  let diff =3D sub pred y in
  mean (mul diff diff)

let result =3D
  (* Choose device - Rune.ocaml, Rune.c=
, Rune.metal *)<=
/span>
  let dev =3D Rune.c in

  (* Initialize parameters on the devic=
e *)
  let w =3D randn dev float32 [| 3; 2 |] in
  let b =3D zeros dev float32 [| 2 |] in
  let x =3D randn dev float32 [| 10; 3 |] in
  let y =3D randn dev float32 [| 10; 2 |] in

  (* Compute loss and gradients<=
span style=3D"color: #8f6f4a; font-style: italic;"> *)
  let loss, grad_w =3D value_and_grad (fun w -> mse_loss ~w ~b x=
 y) w in
  Printf.printf "Loss: %g\n" (unsafe_get [] loss);
  grad_w

val result : (float, Rune.float32_elt, [ `c ]) Rune.t =3D
  [[-1.74967, 0.863766],
   [-0.140407, -0.269364],
   [0.593187, 0.0197736]]
Loss: 2.13033

For more examples and detailed documentation, visit raven-ml.dev.

Why Raven?

Today's machine learning ecosystem is converging on frameworks built atop M= L compilers: high-level Python APIs that build computation graphs, then JIT= compile them for performance.

In parallel, JAX and Flax have gained popularity with th= eir functional APIs. For instance, JAX uses function transformations (grad, jit, vmap) to implements its core features.

This is a landscape where OCaml has natural advantages. OCaml excels at bui= lding compilers, which includes ML compilers, and as a functional language,= it's a more natural fit for functional ML APIs than Python with JAX.

Given these technical advantages, Python's dominance comes down to develope= r experience: the massive ecosystem and excellent prototyping ergonomics. W= e believe that with the right tooling, OCaml can match Python's productivit= y for prototyping and exploratory work. And the ecosystem gap doesn't have = any fundamental challenge: we "just" need to write a lot of code.

If Raven succeeds, we believe it will offer a much more compelling alternat= ive to Python: a language that enables rapid prototyping while eliminating = the gap between exploration and production. You'll move from local developm= ent to production without switching languages, without separate teams, with= out maintaining two stacks.

Technical Highlights: Rune's autodiff engine

One interesting aspect of Raven is Rune's implementation of automatic diffe= rentiation using OCaml's effects system. As far as we know, this is the fir= st production-scale autodiff engine built on effects, drawing on research b= y Jesse Siga= l and earlier work by KC Sivaramakrishn= an.

The architecture follows a modular design:

  • Pluggable backends in Nx allow implementation for different hardware
  • Rune implements an Nx backend that raises effects for all operations.
  • These effects are either caught by an effect handler (e.g. grad, jit), = or, if unhandled, executed eagerly
  • This allows for composable effects handlers (e.g. grad (grad f), jit (g= rad f), etc.)

Technical Highlights: Quill Notebooks

Quill reimagines interactive notebooks with a Typora­like experience. Markdown renders live as you write, sw= itching to raw markdown when you focus on a section for editing. This creat= es a natural writing experience where code blocks integrate naturally into = your document.

The result is a distraction-free notebook environmnet, that prioritizes foc= used writing, while still providing full editor features within code blocks= (coming soon =E2=84=A2!).

While still early, we're excited to see how the community reacts to Quill w= hen it is stable enough for daily use - we really think it has the potentia= l to offer a much better notebook experience for teaching, reading, and oth= er workflows.

Current Status

This is an alpha release. APIs are stabilizing but may still change. Things= will break - this is expected at this stage! If you encounter bugs, please= open an issue on GitHub; community feedback is invaluable to get to a stab= le release.

We're currently focused on:

  • Stabilizing core APIs for the 1.0 release
  • Writing documentation and user guides
  • Supporting early users adoption (FFT, linear algebra, CUDA backend)

Post-Alpha priorities include JIT compilation and stable Quill environment.

Building a Community

One bet we're taking with Raven is that it will allow a scientific and ML c= ommunity in OCaml to flourish. As of now, it's still largely a one-person p= roject. While I'm committed to its development, we really need to see the d= evelopment of a larger community for a project of this size to survive.

If you're interested in the project, the best thing you can do is to engage= =E2=80=94whether by opening issues, reaching out, or contributing. Alongsid= e reaching a first stable release, building a community is Raven's main pri= ority from now on, so any kind of contribution or engagement will be deeply= appreciated. If there's anything I can do to make Raven more welcoming and= approachable, let me know.

I've always believed that the best way to grow OCaml adoption is to provide= killer apps for specific use cases (just like Rails did for Ruby). Raven's= not quite there yet in terms of advantages over Python, but it can get the= re, and if that's something you'd like to contribute to, please reach out\!

Getting Involved

Here are immediate ways to contribute as we work toward a stable release:

For users:

  • Try the libraries with your workflows and report issues
  • Share feedback on API design and usability
  • Help test on different platforms and configurations

For contributors:

  • Optimize eager execution backends while we build JIT compilation
  • Add missing NumPy/Jax/Flax operations to Nx/Rune/Kaun
  • Contribute examples and documentation

Don't hesitate to reach out if you'd like to be involved closely with the c= ore development:

  • JIT compilation
  • Stabilize Quill (many bugs to fix!)
  • New libraries

Resources:

Acknowledgments

I'd like to thank our early contributors and testers who have helped shape = Raven:

  • @axrwl
  • @gabyfle
  • @hesterjeng
  • @ghennequin
  • @kayceesrk
  • @blueavee

Special thanks to our GitHub sponsors for their support:

  • @daemonfire300
  • @gabyfle
  • @sabine

Your feedback, contributions and support have been invaluable in getting Ra= ven to this alpha release - thank you\!

Supporting Raven's Development

As Raven grows, I'm looking for sustainable ways to continue dedicating tim= e to the project. If you're an industrial user interested in using Raven fo= r your machine learning or scientific computing needs, I'd love to talk abo= ut how we can work together.

For individuals who want to support the project, I have a GitHub Sponsors page. Any contribution, = no matter the size, is deeply appreciated and helps ensure Raven's continue= d development.

Slipshow!

Archive: https= ://discuss.ocaml.org/t/ann-slipshow/16337/8

Continuing this thread, Paul-Elliot announced

Let's continue with a new release in this Slipshow thread. It's my ginormous pleasure to announce the opam release of:

Slipshow v0.3.0: The return of the subslips

As you can see, a crucial improvement is that releases are now named. Semve= r is boring, let's add a bit of fun in a serious world.

The release name subtly suggests that subslips are back. Yes, subslips are = back! Slipshow allows your presentation to be visually organized. Subslips = are slips that are inside a slip. You can then "enter" it, go through it, a= nd exit it to go back to the original slip.

But a gif is worth a ginormous word, so here it is:

A demo of slipshow entering subslips

To obtain a similar effect, using the new release, you can simply do: 

We will discuss three topics:

{style=3D"display:flex" children:slip}
----
# Topic 1

Content of the first topic

---
# Topic 2

Content of the second topic

---
# Topic 3

Content of the third topic

If you want to separate the source in multiple file, it's easy: {incl= ude src=3D"file/to/include.md"}. And did you notice that --- now act as group separators? At this point, let me just output the ch= angelog (highligting some of them):

  • Compiler
    • Fix file watching issues by vendoring a (modified) irmin-watcher= , and watching all files the presentation depends on (images, themes, …) (#113)
    • Adds a favicon to the presentation file (Slipshow now has a logo!) (#115)
    • Fix missing attributes on images (#117)
    • Fix missing mime type on images that made svg undisplayable (#120)
    • Fix detection of math inside inline attributes (#124)
    • Add --dimension to specify the dimension of the present= ation (#131)
    • Add less boring name for versions (#132)
  • Language
    • Add {include src=3D"path/to/file.md"} to include a file= in another (#114)
    • Allow pause to have a target (#118)
    • Remove the need for step to execute actions (#118)
    • Added support for subslips and slides (#118)
    • Added pause blocks (#127)
    • Use horizontal lines (---) to group blocks (#129)
    • Pass attributes to children with children: (#130)
    • Consistently remove the need for -at-unpause (#133)
  • Engine
    • Simplify table of content by removing preview (#118)
    • Fix wrong computation of location (#118, #119)
    • Improve zooming behaviour (everywhere) and performance (on chrome-ba= sed browsers…) (#121)
      • If someone has some expertise on how to improve performance on firefox,= I'm interested!
    • Add PageUp and PageDown as navigation keys, adding support for pointers= (#126)
    • Do not act when control is pressed (#126)
    • Fix wrong positioning on scaled slips (#128)
  • Credits

    Thanks to the NLNet foundation for supporting this project!

Other OCaml News

Old CWN

If you happen to miss a CWN, you can send me a message and I'll mail it to you, or go take a loo= k at the archive or the <= a href=3D"https://alan.petitepomme.net/cwn/cwn.rss">RSS feed of the archive= s.

If you also wish to receive it every week by mail, you may subscribe to the= caml-list.

--=-=-=--