On Wed, Aug 20, 2008 at 9:32 AM, Jon Harrop <jon@ffconsultancy.com> wrote:
> On Wednesday 20 August 2008 07:29:21 you wrote:
>
>> 2. Use "dump" files to store the results of these tools and Makefile
>> targets to produce them (that the way ocamldoc is used in Cameleon for
>> browsing documentation)
>
> I have found -dtypes to be very slow on large code bases and I assume this is
> because it generates such huge dump files. Perhaps this could be optimized
> somehow?

Definitely.  .annot files are incredibly bloated.  Here's a snippet
from a .annot file I have on my hard drive:

"""
"ra_live.ml" 342 11092 11107 "ra_live.ml" 342 11092 11117
type(
  Ra_type.code_class
)
"ra_live.ml" 343 11121 11139 "ra_live.ml" 343 11121 11147
type(
  LiveSet.t
)
"ra_live.ml" 344 11148 11172 "ra_live.ml" 344 11148 11177
type(
  Ra_type.label
)
"ra_live.ml" 344 11148 11163 "ra_live.ml" 344 11148 11177
type(
  Ra_type.code_class
)
"""

See any opportunities for optimization?  ;^)

In fact, gzip does a pretty fine job of optimizing .annot files.  The
source file this .annot came from is 15K, the .annot file is 78K, and
gzipping it gets it down to 9K.

Cheers,
-n8

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