This is an adaptation of the original patch submitted in [1], without the
monkey-patching: a patch has been committed in dulwich [2] which allows clients
to supply a custom urllib2 "opener" for opening the url; here, we provide such
an opener, which provides authentication information obtained from the hg
config.
[1] https://groups.google.com/forum/#!topic/hg-git/9clPr1wdtiw
[2] https://bugs.launchpad.net/dulwich/+bug/909037
Consider two octopus merges, one of which is a child of the other. Without this
patch, get_git_parents() called on the second octopus merge checks that each p1
is neither in the middle of an octopus merge nor the end of it. Since the end
of the first octopus merge is a p1 of the second one, this asserts.
Change the sanity check to only make sure that p1 is not in the middle of an
octopus merge.
This was crafted mostly via a bunch of aimless flailing in the
code. I'm pretty well convinced at this point that the incoming
support needs to be rewritten slightly to behave properly in the new
world order (specifically, the overlayrepo class probably should be
subclassing localrepo, or else more directly reimplementing things
instead of trying to forward methods.)
I've been waiting for dulwich upstream to fix this *and* for a test
from domruf that's acceptable. Having gotten neither over a period of
/months/, and having hit the bug myself, I'm moving on and accepting a
patch without tests. This will likely break again, but hopefully
before we'd break it dulwich will be fixed.
Previously, we emitted every Git tree when updating between Mercurial
changesets. With this patch, we now only emit Git trees that changed. A
side-effect of the implementation is that we now only update in-memory
Git trees objects that changed. Before, we always touched Git trees,
invalidating them in the process and causing Dulwich to recalculate
their SHA-1. Profiling revealed this to be expensive and removing the
extra calculation shows a nice performance win.
Another optimization is to not sort the order that changed paths are
processed in. Previously, we sorted by length, longest to shortest.
Profiling revealed that the sorts took a non-trivial amount of time.
While sorted execution resulted in likely idempotent behavior, it
shouldn't be strictly required.
On the author's machine, conversion of the Mercurial repository itself
decreased from ~493s to ~333s. Even more impressive is conversion of
Firefox's main repository (which is considerably larger). Converting the
first 200 revisions of that repository decreased from ~152s to ~42s.
This replaces the brute force Mercurial to Git export with one that is
incremental. It results in a decent performance win and paves the road
for parallel export via using multiple incremental exporters.
A recent real world occurrence - user hand edited the timezone field in
an hg export to provide a unique value (from prior export). Hg imported
the export okay, but dulwich threw an exception.
This test shows the fault.
If dulwich is presented with a "sub minute" timezone offset, it throws
an exception (see tests/test-timezone.t). This patch rounds the timezone
down to the next minute before passing the value to dulwich.
As pointed out by l33t, Hg-Git's output for push doesn't currently do a very
good job of telling the user what happened. My previous changes in this area
had moved some of the output from status to note, making it only show if
--verbose was specified. However, I hadn't realized at the time that the
reference information (though overly verbose) was providing a valueable purpose
that otherwise wasn't met; telling the user that a remote reference had changed.
This changeset makes it so that:
* default output will include simple messages like "adding reference
refs/heads/feature" and "updating reference refs/heads/master" (omitting any
mention of unchanged references)
* verbose output will include more detailed messages like "adding reference
default::refs/heads/feature => GIT:aba43c" and "updating reference
default::refs/heads/master => GIT:aba43c" (omitting any mention of unchanged
references)
* debug output will include the detailed output like in verbose, but
addtionally will include messages like "unchanged reference
default::refs/heads/other => GIT:aba43c"
https://bitbucket.org/durin42/hg-git/issue/64/push-confirmation
l33t pointed out that currently, Hg-Git doesn't provide any confirmation that a
push was successful other than the exit code. Normal Mercurial provides a
couple other messages followed by "added X changesets with Y changes to
Z files". After this change, Hg-Git will provide much more similar output.
It's not identical, as the underlying model is substantially different, but the
concept is the same. The main message is "added X commits with Y trees and
Z blobs".
This change doesn't affect the output of what references/branches were touched.
That will be addressed in a subsequent commit.
Dulwich doesn't provide an easy hook to get the information needed for this
output. Instead of passing generate_pack_contents as the pack generator
function to send_pack, I pass a custom function that determines the "missing"
objects, stores the counts, and then calls generate_pack_contents (which then
will determine the "missing" objects again.
The new expected output:
searching for changes # unless quiet true
<N> commits found # if verbose true
list of commits: # if debugflag true and at least one commit found
<each hash> # if debugflag true and at least one commit found
adding objects # if at least one commit found unless quiet true
added <N> commits with <N> trees and <N> blobs # if at least one object unless
# quiet true
https://bitbucket.org/durin42/hg-git/issue/64/push-confirmation
In f32e473ff520, the "commit" function was extracted into a testutil for re-use.
However, test-encoding.t was skipped over in that changeset, as I was seeing
unexplained test failures. Since those test failures have now been explained
(and fixed), this changeset performs the same extraction on test-encoding.t as
was done on all the other tests.
The version of fn_git_commit that was used in testutil redirected all output
(including errors) to /dev/null, which didn't match the expectations of this
test. The test utility functions for commit/tag now no longer throw away error
output, instead leaving it to individual tests to decide if error output should
be ignored.
It looks like Git 1.8.0 started silently converting latin1 commit messages to
utf-8. That changed the result of this test. This changeset alters the test
to make it accept both the pre-1.8.0 and post-1.8.0 behaviors.
https://raw.github.com/git/git/master/Documentation/RelNotes/1.8.0.txt