hgsubversion Google Group

[hgsubversion] push by durin42 - revsets: fix nativerevs fallback on 2013-05-18 14:34 GMT

hgsubvers...@googlecode.com — Sat, 18 May 2013 14:34:38 UT

Revision: f0ebc7113429
Branch: default
Author: Matt Mackall <m...@selenic.com>
Log: revsets: fix nativerevs fallback
[link]

Modified:
/hgsubversion/util.py

Re: [PATCH] revsets: fix nativerevs fallback

r...@durin42.com (Augie Fackler) — Sat, 18 May 2013 14:34:19 UT

Thanks, pushed

Re: [PATCH 6 of 6 RESEND] layouts: use layouts library for branch mappingduring push

r...@durin42.com (Augie Fackler) — Sat, 18 May 2013 14:29:23 UT

Pending the __unimplemented response, series LGTM.

Re: [PATCH 1 of 6 RESEND] layouts: refactor out svn path to mercurial branch logic

r...@durin42.com (Augie Fackler) — Sat, 18 May 2013 14:28:04 UT

I think we've had this conversation before, but why this instead of raise NotImplementedError?

[PATCH 1 of 6 RESEND] layouts: refactor out svn path to mercurial branch logic

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:32 UT

layouts: refactor out svn path to mercurial branch logic

This pulls the logic for mapping from svn path to mercurial branch
name out of svnmeta.py and into the new layouts library. It also sets
up the structure for that library. This diff does not modify any call
to svnmeta.localname, rather leaving it in place as a simple proxy to

[PATCH 3 of 6 RESEND] layouts: refactor path component of convinfo generation into layouts lib

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:34 UT

layouts: refactor path component of convinfo generation into layouts lib

This adds code responsible for producing an absolute path within the
subversion repository to the new layout objects. It also uses that
code to generate the path component of the conversion info we stick
into extra.

diff --git a/hgsubversion/layouts/base.py b/hgsubversion/layouts/base.py

[PATCH 4 of 6 RESEND] layouts: use layouts library for branch mapping with generating .hgignore

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:35 UT

layouts: use layouts library for branch mapping with generating .hgignore

diff --git a/hgsubversion/svncommands.py b/hgsubversion/svncommands.py
--- a/hgsubversion/svncommands.py
+++ b/hgsubversion/svncommands.py
@@ -374,10 +374,9 @@
hashes = meta.revmap.hashes()
parent = util.parentrev(ui, repo, meta, hashes)

[PATCH 5 of 6 RESEND] layouts: use layout library in hg svn info

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:36 UT

layouts: use layout library in hg svn info

diff --git a/hgsubversion/svncommands.py b/hgsubversion/svncommands.py
--- a/hgsubversion/svncommands.py
+++ b/hgsubversion/svncommands.py
@@ -413,17 +413,8 @@
return 0
r, br = hashes[pn]
subdir = util.getsvnrev(parent)[40:].sp lit('@')[0]

[PATCH 6 of 6 RESEND] layouts: use layouts library for branch mappingduring push

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:37 UT

layouts: use layouts library for branch mappingduring push

diff --git a/hgsubversion/pushmod.py b/hgsubversion/pushmod.py
--- a/hgsubversion/pushmod.py
+++ b/hgsubversion/pushmod.py
@@ -99,12 +99,7 @@
file_data = {}
parent = rev_ctx.parents()[0]
parent_branch = rev_ctx.parents()[0].branch()

[PATCH 2 of 6 RESEND] layouts: refactor mercurial branch to svn path mapping out of svnmeta

dschlei...@fb.com (David Schleimer) — Fri, 17 May 2013 18:48:33 UT

layouts: refactor mercurial branch to svn path mapping out of svnmeta

This leaves the existing svnmeta.remotename as a proxy to the new
layout objects rather than updating any of the callsites.

diff --git a/hgsubversion/layouts/base.py b/hgsubversion/layouts/base.py
--- a/hgsubversion/layouts/base.py

[PATCH] revsets: fix nativerevs fallback

m...@selenic.com (Matt Mackall) — Tue, 14 May 2013 18:13:48 UT

revsets: fix nativerevs fallback

diff -r f0bde4bf1f78 -r f0ebc7113429 hgsubversion/util.py
--- a/hgsubversion/util.py Tue Apr 23 13:59:38 2013 -0500
+++ b/hgsubversion/util.py Fri May 10 18:52:28 2013 -0500
@@ -343,7 +343,7 @@
def revset_stringset(orig, repo, subset, x):
if x.startswith('r') and x[1:].isdigit():

[PATCH] updatemeta/rebuildmeta: handle hidden changesets

d...@cabo.dk (Dan Villiom Podlaski Christiansen) — Mon, 13 May 2013 09:53:18 UT

updatemeta/rebuildmeta: handle hidden changesets

diff --git a/hgsubversion/svncommands.py b/hgsubversion/svncommands.py
--- a/hgsubversion/svncommands.py
+++ b/hgsubversion/svncommands.py
@@ -135,7 +135,12 @@ def _buildmeta(ui, repo, args, partial=F
# changesets that close a branch, and store their first parent

Re: svn author lookup case insensitive?

r...@durin42.com (Augie Fackler) — Wed, 08 May 2013 15:05:58 UT

Why not just .lower() authors on the way into the author map and then again when doing the membership check two lines earlier? That'd make the code clearer and more correct.

Also, a test would be nice.

Re: svn author lookup case insensitive?

augustin.mich...@googlemail.com (maugustin) — Sun, 05 May 2013 17:51:48 UT

Ok. Here is the patch. I can't use hg email since I'm working in a MS
Outlook environment. There's no SMTP server arround and only HTTP-proxy
internet access.
The patch:
additional authormap lookup with lower-case username
diff --git a/hgsubversion/maps.py b/hgsubversion/maps.py
--- a/hgsubversion/maps.py

Re: svn author lookup case insensitive?

r...@durin42.com (Augie Fackler) — Fri, 03 May 2013 13:57:50 UT

On Fri, May 3, 2013 at 3:14 AM, maugustin

Okay, given that, I'd look at a patch. Send a patch inline, so I can
comment on it in my mailreader (hg email does the right thing by
default, and is generally good about not mangling patch contents).

Re: svn author lookup case insensitive?

augustin.mich...@googlemail.com (maugustin) — Fri, 03 May 2013 07:14:02 UT

Hi Augie,
i've checked login with option --username and --no-auth-cache and I could
login with username written in lowercase and also in uppercase letters.
This worked on Windows 7 and on Ubuntu Linux clients. So I think SVN is not
case-sensitive on usernames.
I think it makes the authorname replacement more stable.

Re: svn author lookup case insensitive?

r...@durin42.com (Augie Fackler) — Thu, 02 May 2013 13:28:45 UT

I could be mistaken, but I don't think Subversion treats usernames as
case-insensitive, I think that might be a Windows defect?

On Thu, May 2, 2013 at 4:03 AM, maugustin

svn author lookup case insensitive?

augustin.mich...@googlemail.com (maugustin) — Thu, 02 May 2013 08:03:00 UT

Hi developer,
would it make sense to do the author lookup during SVN pull run case
insensitive?
We migrated some maschines from Windows XP to Windows 7 and now the
username from Windows is not lowercase but uppercase. I've put up now every
entry in authormap file twice (additional with all letters uppercase) and

[hgsubversion] push by durin42 - revsets: add support for 'r123' revision labels... on 2013-04-30 17:15 GMT

hgsubvers...@googlecode.com — Tue, 30 Apr 2013 17:15:40 UT

Revision: f0bde4bf1f78
Branch: default
Author: Matt Mackall <m...@selenic.com>
Log: revsets: add support for 'r123' revision labels

This is controlled by an hgsubversion.nativerevs boolean (off by default).
[link]

Modified:

Re: [PATCH] revsets: add support for 'r123' revision labels

r...@durin42.com (Augie Fackler) — Tue, 30 Apr 2013 16:41:50 UT

queued, thanks

[PATCH 6 of 6] layouts: use layouts library for branch mappingduring push

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:12 UT

layouts: use layouts library for branch mappingduring push

[PATCH 5 of 6] layouts: use layout library in hg svn info

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:11 UT

layouts: use layout library in hg svn info

[PATCH 4 of 6] layouts: use layouts library for branch mapping with generating .hgignore

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:10 UT

layouts: use layouts library for branch mapping with generating .hgignore

[PATCH 3 of 6] layouts: refactor path component of convinfo generation into layouts lib

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:09 UT

layouts: refactor path component of convinfo generation into layouts lib

diff --git a/hgsubversion/layouts/base.py b/hgsubversion/layouts/base.py

[PATCH 2 of 6] layouts: refactor mercurial branch to svn path mapping out of svnmeta

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:08 UT

layouts: refactor mercurial branch to svn path mapping out of svnmeta

This leaves the existing svnmeta.remotename as a proxy to the new
layout objects rather than updating any of the callsites.

diff --git a/hgsubversion/layouts/base.py b/hgsubversion/layouts/base.py
--- a/hgsubversion/layouts/base.py

[PATCH 1 of 6] layouts: refactor out svn path to mercurial branch logic

dschlei...@fb.com (David Schleimer) — Mon, 29 Apr 2013 23:10:07 UT

layouts: refactor out svn path to mercurial branch logic

[PATCH] revsets: add support for 'r123' revision labels

m...@selenic.com (Matt Mackall) — Wed, 24 Apr 2013 02:32:59 UT

revsets: add support for 'r123' revision labels

This is controlled by an hgsubversion.nativerevs boolean (off by default).

diff -r bdc9b21ea8d0 -r f0bde4bf1f78 hgsubversion/__init__.py
--- a/hgsubversion/__init__.py Mon Apr 22 14:27:05 2013 -0700
+++ b/hgsubversion/__init__.py Tue Apr 23 13:59:38 2013 -0500

Re: svn-style revision ids

ole...@gmail.com (Olemis Lang) — Tue, 23 Apr 2013 20:08:44 UT

[...]

Count me in . I submit hundreds of patches to Apache™ Bloodhound
developed on a MQ patch queue , so referencing svn changesets is
something I do many times every day .

PS: BTW , git-svn seems to support r123 changeset IDs , though that's
something bothering me on a regular basis ;)

Re: svn-style revision ids

ole...@gmail.com (Olemis Lang) — Tue, 23 Apr 2013 20:03:36 UT

[...]

On svn , yes , as well as any other version control system defining a
changeset ID 123 (in the case of r123)

[...]

--
Regards,

Olemis.

Apache™ Bloodhound contributor
[link]

Blog ES: [link]
Blog EN: [link]

Re: svn-style revision ids

m...@selenic.com (Matt Mackall) — Tue, 23 Apr 2013 18:29:58 UT

No, that should be fine.

--
Mathematics is the supreme nostalgia of our time.

Re: svn-style revision ids

r...@durin42.com (Augie Fackler) — Tue, 23 Apr 2013 17:31:45 UT

I guess I'm happy with this in an off-by-default capacity? I assume you're already pushing a config for hg to all user machines anyway. We can always turn it on by default later. Anyone object to that?

(I'm not planning on writing this, but will welcome the patch.)

Re: svn-style revision ids

b...@serpentine.com (Bryan O'Sullivan) — Tue, 23 Apr 2013 17:29:12 UT

I'm very much in favour of Matt's suggestion.

Re: svn-style revision ids

m...@selenic.com (Matt Mackall) — Tue, 23 Apr 2013 16:33:59 UT

That argument doesn't make any sense to me. Yes, I can see an argument
against namespace pollution. I can't see an argument for avoiding said
pollution.. just because another extension might hypothetically want to
do it.

As for Trac in particular, if we're in an environment that is based on

Re: svn-style revision ids

r...@durin42.com (Augie Fackler) — Tue, 23 Apr 2013 15:50:13 UT

I guess the difficulty would be if we squat on the rXXX nomenclature and then some other extension wants to make that work for (e.g.) trac compatibility or something.

I'm -0 on just r, but would be happy with svnXXX as an operator, I guess. Is this mostly to avoid shell quoting on the svnrev() set?

[hgsubversion] push by durin42 - setup: add layouts subdir on 2013-04-23 15:45 GMT

hgsubvers...@googlecode.com — Tue, 23 Apr 2013 15:45:48 UT

Revision: bdc9b21ea8d0
Branch: default
Author: Bryan O'Sullivan <bry...@fb.com>
Log: setup: add layouts subdir
[link]

Modified:
/setup.py

Fwd: svn-style revision ids

ole...@gmail.com (Olemis Lang) — Sun, 21 Apr 2013 15:42:13 UT

The fact is that , in general , hgsubversion might not be the only
tool operating on a target repository . OTOH for e.g. Trac +
Bloodhound (... and maybe others ? ... ) r123 is a valid TracLinks
expression for any kind of repository .

AFAICT they are *virtual* tags , in the following sense

Re: svn-style revision ids

timothy.c.dela...@gmail.com (Tim) — Sat, 20 Apr 2013 23:02:21 UT

The way I've worked around this is to override
svnrepo.svnlocalrepo.nodetags( ) to inject the svn tags (format I use is is
'svn/123') - the name appears as a tag when requesting info about the node,
but doesn't appear in the repo's list of tags. And to lookup a tag, I
override svnreop.svnlocalrepo.__getitem __. Augie previously rejected

Re: svn-style revision ids

m...@selenic.com (Matt Mackall) — Sat, 20 Apr 2013 16:48:15 UT

I may be mistaken that r123 is standard nomenclature in SVN-land, as
I've carefully avoided using it. But to quote a random snippet of the
red bean book:

$ svn checkout -r 1729 # Checks out a new working copy at r1729
…
$ svn update -r 1729 # Updates an existing working copy to r1729
…

Re: svn-style revision ids

m...@selenic.com (Matt Mackall) — Sat, 20 Apr 2013 16:20:14 UT

Probably the simplest thing is to wrap revset.stringset.

--
Mathematics is the supreme nostalgia of our time.

Re: svn-style revision ids

ole...@gmail.com (Olemis Lang) — Sat, 20 Apr 2013 00:23:51 UT

Hi !

I'll try to briefly express my opinion . r123 ids are appropriate for
systems offering an abstract repository API (e.g. Trac , Bloodhound ,
JIRA , issue trackers , CI build tools , ...) , but IMHO not in this
case . The r (for revision AFAICT) does not offer any hints about the
target repos type . I'd rather prefer an style similar to HgSvnUtils

svn-style revision ids

r...@durin42.com (Augie Fackler) — Fri, 19 Apr 2013 23:45:39 UT

Interesting thought. I think I did this years ago with a horrible hack involving tags, but it always rubbed the the wrong way. Do you have something in particular in mind? The explicitness of the revset operator felt "right".

Re: Can not clone svn repository needing http authorization?

bond...@gmail.com — Fri, 19 Apr 2013 20:27:25 UT

Thank you! This did the trick for me.

[hgsubversion] 7 new revisions pushed by durin42 on 2013-04-18 22:28 GMT

hgsubvers...@googlecode.com — Thu, 18 Apr 2013 22:28:37 UT

7 new revisions:

Revision: 26aa0172725b
Branch: default
Author: Augie Fackler <r...@durin42.com>
Log: test_util: specify a username in our test-run hgrc.
[link]

Revision: 117b3b421294
Branch: default
Author: David Schleimer <dschlei...@fb.com>

[PATCH 4 of 4] layouts: refactor buildmeta layout reading and writing

dschlei...@fb.com (David Schleimer) — Thu, 18 Apr 2013 20:18:22 UT

layouts: refactor buildmeta layout reading and writing

diff --git a/hgsubversion/layouts/detect. py b/hgsubversion/layouts/detect. py
--- a/hgsubversion/layouts/detect. py
+++ b/hgsubversion/layouts/detect. py
@@ -70,3 +70,18 @@
if ui:
ui.setconfig('hgsubversion', 'layout', layout)

[PATCH 3 of 4] layouts: refactor layout loading and persisting out of svnmeta.py

dschlei...@fb.com (David Schleimer) — Thu, 18 Apr 2013 20:18:21 UT

layouts: refactor layout loading and persisting out of svnmeta.py

diff --git a/hgsubversion/layouts/__init_ _.py b/hgsubversion/layouts/__init_ _.py
--- a/hgsubversion/layouts/__init_ _.py
+++ b/hgsubversion/layouts/__init_ _.py
@@ -10,7 +10,9 @@
"""

import detect
+import persist

__all__ = [

[PATCH 2 of 4] layouts: refactor layout loading based on config into function

dschlei...@fb.com (David Schleimer) — Thu, 18 Apr 2013 20:18:20 UT

layouts: refactor layout loading based on config into function

Refactor the logic for loading the layout based on config, and for
validating that config is sane into a single function.

diff --git a/hgsubversion/layouts/detect. py b/hgsubversion/layouts/detect. py
--- a/hgsubversion/layouts/detect. py

[PATCH 1 of 4] layouts: pull out code for detecting layout from subversion

dschlei...@fb.com (David Schleimer) — Thu, 18 Apr 2013 20:18:19 UT

layouts: pull out code for detecting layout from subversion

This is the start of an effort to refactor the code for how we handle
subversion layoout to mercurial layout mapping. Specifically, it
pulls out the logic for guessing what layout to use based on the
contents of subversion. It also sets up a bit of the structure for

[PATCH] testutil: clear rcpath when we update HGRCPATH env var

dschlei...@fb.com (David Schleimer) — Fri, 12 Apr 2013 22:59:36 UT

testutil: clear rcpath when we update HGRCPATH env var

hg maintains a global cache of the HGRCPATH that doesn't get updated
when HGRCPATH is changed. Nuke said global cache when we change the
HGRCPATH.

diff --git a/tests/test_util.py b/tests/test_util.py
--- a/tests/test_util.py
+++ b/tests/test_util.py

[PATCH] buildmeta: verify uuid when passed explicit url

dschlei...@fb.com (David Schleimer) — Fri, 12 Apr 2013 18:42:15 UT

buildmeta: verify uuid when passed explicit url

2c87bdc43d3c changed buildmeta to read the uuid from local disk, if
it's available. As a side effect, it disabled the validation of the
uuid we've recorded locall in the commits against the uuid of the repo
we are building against. This is probably reasonable when someone

no_heads.patch

cam...@gmail.com (David Sankel) — Fri, 05 Apr 2013 02:03:56 UT

This patch allows hgsubversion to be used as a guest repository with the
guestrepo extension. The pull request that this patch is based on was
rejected with this comment:
Please resend this as a patch to the hgsubversion mailing list. I'm kind of
opposed to it, but I'm not sure I have a solid reason why, it's just a gut