[ViewCVS] Re: [Zope3-dev] svn.zope.org tarballs?

[C. Michael Pilato]

Jim Fulton <jim at zope.com> writes:

> I finally did some checking and found that downloading a tar ball
> causes ViewCVS to use massive amounts of memory. For example, to
> download a 3MB tar ball of the Zope 3 trunk, ViewCVS consumed 412 MB
> of memory.

This one's for you, Jim.  Let me know if it helps out -- it made a
*huge* difference for me.

<<<
Fix a bunch of pool usage problems with ViewCVS/Subversion.

* viewcvs/lib/vclib/svn/__init__.py
  (_fetch_log): Now accept a 'pool' parameter.
  (FileContentsPipe): Was StreamPipe().
  (FileContentsPipe.__init__): Instead of taking a stream, accept the
    root, path, and a pool.  Use a subpool of the pool for the stream
    creation.
  (FileContentsPipe.__del__): New.
  (SubversionRepository.__init__): Make a scratch subpool that
    operations can safely use and clear.
  (SubversionRepository._scratch_clear): New convenience function.
  (SubversionRepository.itemtype, SubversionRepository.listdir): Use
    and clear the scratch pool.
  (SubversionRepository.filelog): Update call to _fetch_logs().  Use
    and clear the scratch pool.
  (SubversionRepository.openfile): Update call to FileContentsPipe().
    Use and clear the scratch pool for the call to _get_last_history_rev().
>>>

Checking in lib/vclib/svn/__init__.py;
/cvsroot/viewcvs/viewcvs/lib/vclib/svn/__init__.py,v  <--  __init__.py
new revision: 1.55; previous revision: 1.54
done