Re: [Caml-list] New Ocaml Plug-in for NetBeans

["Nathaniel Gray" <n8gray@gmail.com>]

On Tue, Sep 9, 2008 at 12:43 AM, Jon Harrop
<jonathandeanharrop@googlemail.com> wrote:
> On Tuesday 09 September 2008 06:31:51 you wrote:
>> Sure, I had the same thought, but gzip already exists, doesn't require
>> approval from INRIA, and .gz is easily read in just about any language
>> (including OCaml).  Once the .annot files are smaller than the source
>> files I'm completely willing to generate them all the time.
>
> I think you will find that GZip is orders of magnitude less efficient that the
> approach I suggested and too slow to be useful in this case.
>
> For example, gzipping and unzipping the .annot files from the 4kLOC core of
> Smoke takes 0.5s. So GZip is too slow for interactive use even on this tiny
> code base.

Why would you want to unzip more than one .annot file during
interactive use?  If I query the type of some token foo in bar.ml I
only need to read from bar.annot.gz, not all the files in the project.
 The biggest .annot file I have lying around is from a 40KB source
file.  It's about 270KB before gzip, 30KB after.  According to "time"
it takes 0.009 seconds to gunzip on my MacBook Pro.  I'd say that's
plenty fast enough to be useful, even if your machine isn't quite as
snappy as mine.

I don't dispute that a more efficient encoding than gzip can be found,
but good luck convincing anybody at INRIA to work on it.  ;^)

-- 
>>>-- Nathaniel Gray -- Caltech Computer Science ------>
>>>-- Mojave Project -- http://mojave.cs.caltech.edu -->