Error - view large changeset

Issue #129 resolved
Peter Miller
created an issue

When attempting to view a large changeset (000s of files) of a small repository (5 commits), a 500 error is given, details of which are:

{{{ Module rhodecode.controllers.changeset:86 in index

if filenode_old.is_binary or node.is_binary: Module vcs.utils.lazy:26 in get result = obj.dict[self.name] = self._func(obj) Module vcs.nodes:342 in is_binary return bool(self.content and '\0' in self.content) Module vcs.utils.lazy:26 in get result = obj.dict[self.name] = self._func(obj) Module vcs.nodes:240 in content return safe_unicode(content) Module vcs.utils:33 in safe_unicode u_str = unicode(str) MemoryError:

}}}

This is repeatable in that it happens every time this changeset is attempted to be viewed. Other changesets (1 file) in the same repo are successfully viewed. Is there a way to fix this?

Thanks again,

Pete

PS: there is no version selection for 1.1.4 in this issue tracker, but I believe this is the latest version, and is the version I am using.

Comments (3)

  1. Marcin Kuzminski repo owner
    • changed milestone to 1.1

    First You can try changing vcs to version 0.1.1 this version have fixed some memory problems with vcs. Maybe this will help.

    Secondly most of this kind of issues are fixed in 1.2 beta branch (+ vcs 0.2 beta). Maybe You can try to install 1.2 and test it. (works fine for openbsd repo 2gb 150K+ commits)

  2. Marcin Kuzminski repo owner

    I did many improvements in latest tip of rhodecode and vcs, also I made some tests on artificial repos containing a lot of binary files, i things look much better now. Can You test Your use case on the problematic repository ?

  3. Log in to comment