I spent a little time review Mercurial's open issues list
(http://www.selenic.com/mercurial/bts/), to see if there is anything
that we need to have fixed before we move ON to Mercurial.  After some
discussion, Steve, Rich, and I have a draft list of showstopper bugs,
and another list of non-stoppers that seemed important enough to at
least mention.

The lists are in the attached HTML file.  References to "cdm" or
"cadmium" refer to an extension that Rich is working on that will
give us certain bits of wx functionality (e.g., delta squishing and
backup of modified source files).

We're still trying to get a handle on rename functionality.  Basic
rename support is in Mercurial 0.9.3, but there are several edge cases
that we're not sure about.  One of the things I want to do in the next
week or so is review the existing test suite and see if there are any
rename tests that we want to add.

As far as tracking these issues, I was thinking of opening shadow bugs
on bugs.grommit.com, rather than add Solaris-specific keywords to the
Mercurial bug database.  If anyone can think of a better approach,
please let me know.

If you have any comments on the lists, especially items that should be
moved from one list to the other, please speak up.

thanks,
mike
Title: Draft Migration Stoppers (Mercurial Issues)

Draft Migration Stoppers (Mercurial Issues)

Link to as http://www.selenic.com/mercurial/bts/issueNNN.

issue# synopsis notes
132 hg should revalidate its data after locking the repo Can corrupt repo? Discuss with mpm?
455 Mismerge involving two concurrently renamed directories
498 unreliable tag computation when tags change over time a patch exists?
199 detect corrupted dirstate
274 localrepo.add() fails to unlock
222 [extensions] in a repo's .hg/hgrc file do not get loaded appears to make it harder to use ON-specific extensions only in ON repos.
226 Strange hgmerge invocation behaviour could lead to mismerges.
122 commit succeeds if merge fails makes mismerges easier than with Teamware.
269 hg mv should handle newly added files we've already bumped into it a few times during testing
401 Problem with "hg rm -A" after "hg mv" has workaround. very similar to 269.
413 hg add mishandles subprojects possible repo corruption

While these issues are potentially serious, we're not considering them stopper candidates:

issue# synopsis notes
273 dirstate.update is racy only triggered if a hook updates the working directory when it shouldn't
312 revert clobbers .orig files should be able to work around
29 merging directories and files is unhelpful unlikely to occur in practice; should be able to work around
387 Closing heads during push causes exit code 1 gate shouldn't have multiple heads in the first place. (resolved but not in 0.9.3?)
429 notify hook fails after upgrade looks like the fix is in 0.9.3 (based on the timeline). If not and it's a serious problem, we'll run into it during testing.
451 Strip with multiple heads acts strangely from the issue notes, it looks like the fix to prevent the repo from getting messed up is in 0.9.3. however, the command is still called "strip" instead of "qstrip" in 0.9.3.
366 Commit misses changes during commit has potential for introducing mismerge, but seems unlikely to happen in practice.
382 hg status shows too many files modified, fixed on the next call bug in underlying platform, rather than Mercurial? NFS attribute caching issue?
337 grep -all does not work as advertised. for finding when text was added or removed, we can work around by using "hg bisect". Fixed? Released?
136 Mercurial misses changes for files with close creation times hard to hit
93 hg revert should mark rename files in the opposite direction shouldn't come up much; has workaround
456 Diff doesn't follow renames arguably not a bug, since standard diff format doesn't support renames. Not an issue for webrev. See also issue295.
470 Inconsistent update behaviour for modified files when going in the past only causes problems if there are modified files in the working copy, which shouldn't happen for things like gate backouts
236 running out of disk space behaves weirdly doesn't corrupt repo
8 need to show rename info will get this from cdm
507 hg merge -f doesn't work after a failed merge makes merging significantly more brittle than with Teamware, but probably not an issue as long as most people use a merge tool that produces a sane exit code.

AFAIK Teamware doesn't do anything with character encodings--whatever encoding was used when the file was created is the encoding you get. So I'm going to ignore all the encoding issues.


@(#)hg-stoppers.html 1.2 2007/04/23 15:07:40 (kupfer)
_______________________________________________
tools-discuss mailing list
[email protected]

Reply via email to