Mercurial test conversion

[Felipe Contreras]

On Sun, May 27, 2012 at 5:07 PM, John Bailey <rekkanoryo at rekkanoryo.org> wrote:

> I have completed a test conversion of our monotone database to mercurial.  I ran
> the conversion on Friday evening, but it took most of yesterday to push the
> resulting repositories to a location that was publicly accessible.  Here are the
> repositories (all accessible via any browser or via 'hg clone' with the URL's
> listed):
>
> This repository is the raw result from the 'hg convert' run.  It includes
> everything currently in our monotone database:
>
> http://hg.guifications.org/pidgin-test-2012-05-24/convert_result

It seems you didn't apply the 'hg convert' patches:
http://hg.guifications.org/pidgin-mtn-conv-files/file/1ca362d9b659/hg-convert-fix-rev-in-changelog.diff
http://hg.guifications.org/pidgin-mtn-conv-files/file/1ca362d9b659/hg-convert-handle-mtn-comments-and-committers.diff

So the committer is missing, and comments, and mtn ids are not
converted to hg ids.

In addition to that, you don't have the extra mtn information that
last I checked Richard Laager was interested in keeping (like mtn id).

Moreover, you merged my 'test' bookmark which wasn't supposed to be
merged, and now the scripts aren't actually doing any conversion.

Also, since the last commit was 16 months ago, and the last tag is
v2.7.11[1], I don't think this is a proper conversion at all.

> Additional notes:  I reviewed and accepted the conversion script work Felipe
> Contreras submitted to the list back on 2012-05-13.

You merged unintended changes.

I don't see any other changes in pidgin-mtn-conv-files than mine, and
last time I checked those scripts were definitely not ready for
prime-time.

At the very least I see this list of unmapped authors:

tim at retout.co.uk | 1
thijs at adium.im | 1
pat at erley.org | 1
loentar at google.com | 1
Eion Robb <eion at robbmob.com> | 1
Patrick Cloke | 1
Simon van der Linden <simon at vanderlinden.eu.org> | 1
lelongt at gmail.com | 1
jond578 at gmail.com | 1
dzlists at arcor.de | 1
emanuele.giaquinta at gmail.com | 1
matthew at bells23.org.uk | 1
jamie at lacewing-project.org | 1
dvpdiner2 at gmail.com | 1
dcbw at redhat.com | 2
linus.luessing at web.de | 3
olivier.crete at collabora.com | 3
hyc at symas.com | 5
michael at soc.pidgin.im | 18
sanket at soc.pidgin.im | 22
sanket at sanketagarwal.com | 26
tomkiewicz at cpw.pidgin.im | 64

Many should be fixed.

Also, there is the topic of author mapping convention. Richard Laager
mentioned something like:

john <john at snow.com>
John Snow <unknown>

That can be easily done with 'mtn git_export'.

At this point in time I think the mtn -> git -> hg conversion is in
better shape.

 1) monotone information

   Monotone-Parent: 9e9d09486a04f5adb2e51b0f5aada69a9b053e93
   Monotone-Revision: 2cca525615a367389b1639a5246fbe6d451150c4

   Monotone-Author: markdoliner at pidgin.im
   Monotone-Date: 2012-02-18T23:28:43
   Monotone-Branch: im.pidgin.pidgin.2.x.y
   Monotone-Branch: im.pidgin.pidgin.mxit.2.x.y

 2) Multiple changelogs are maintained

 3) Proper default author mappings

 4) Much faster

Right now both methods require patching mercurial extensions, but when
I'm done with mine that won't be needed.

Cheers.

[1] http://hg.guifications.org/pidgin-test-2012-05-24/convert_result/rev/e3286024d057

-- 
Felipe Contreras