475012 - Create an own hg repository for Venkman

Status: VERIFIED FIXED

(Other Applications Graveyard :: Venkman JS Debugger, defect)

Description

[Serge Gautherie (:sgautherie)]

+++ This bug was initially created as a clone of Bug #437064 +++

Or at least a mirror (at http://hg.mozilla.org/cvs-trunk-mirror/file/default/extensions/ or elsewhere)

Comment 1

[:Gijs (he/him)]

Same question as for bug 475012: why?

You don't just tell people "do this work" without at least including a tiny clue about why it would be useful.

Comment 2

[Serge Gautherie (:sgautherie)]

See my answer at bug 475011 comment 2.

Sorry, I (wrongly) thought the current situation was quite obvious.

Comment 3

[:Gijs (he/him)]

For Venkman, the situation is a bit different. In fact, almost nobody works on this code anymore. If moving to HG changes this, then for all I care we can move. However, I severely doubt it is worth the effort. If venkman is to survive, more people need to have time to contribute. I certainly don't, and I haven't heard about many people who do. I try to review stuff whenever it comes up, and I believe that so far, that is OK. But I don't have time to code stuff, and given the number of regressions that have occurred, we really need someone to take us back up where we should be in terms of how well things work. Anyway, I may try to do something about this at FOSDEM. We'll see.

Comment 4

[Karsten Düsterloh]

(In reply to comment #3)
> In fact, almost nobody works on this code anymore.

Sad, but true. 
OTOH, many important bugs depend upon core JS engine code being fixed, which doesn't quite allow for "drive-by patches".

> If moving to HG changes this, then for all I care we can move.

Hard to tell. 
At least localizers would be pleased to have one VCS less (once all extensions in CVS would move as well). ;-)

> However, I severely doubt it is worth the effort. If venkman is to
> survive, more people need to have time to contribute.

The need to set up CVS just for Venkman isn't particularly motivating for non-SeaMonkey Mozilla devs, I'd suppose...

> I certainly don't, and I haven't heard about many people who do.

We should file a bug on "Days are too short". :-(

> Anyway, I may try to do something about this at FOSDEM. We'll see.

I'll be there, too.

Comment 5

[Serge Gautherie (:sgautherie)]

(In reply to comment #3)
> For Venkman, the situation is a bit different. In fact, almost nobody works on

Indeed, yet: what eventually made me file these bugs is bug 473936 (for example), which includes a patch for Venkman...

> this code anymore. If moving to HG changes this, then for all I care we can

I know (a little) ... to make it short:
see my answer at bug 475011 comment 8:
a mirror would be just fine to start with;
unless you'll want to go for a full migration already.

> Anyway, I may try to do something about this at FOSDEM. We'll see.

(I'll be there, too.)

Comment 6

[Alex Vincent [:WeirdAl]]

Well, I can come up with one good reason why a hg repo would be a good idea:  we run the risk of introducing a change on cvs that means it won't compile or execute with 1.9 code (the rest of CVS) but will compile and execute on 1.9.1+ code.

The CVS repo is for all intents and purposes a stability branch now.  Trunk/1.9.1 development should not happen on it, even for NPOTB code like Venkman.

Comment 7

[:Gijs (he/him)]

Right. For Venkman, I would be fine with an actual HG repo. There is less pain involved also because fewer people actually care about the CVS repo that we currently have, and more of the people who use HG repo care about Venkman. With ChatZilla, things are the other way around.

Anyway, I have no idea how this would actually work. KaiRo?

Comment 8

[Robert Kaiser]

Do it the same way as it was done for DOMi - see bug 437064 and bug 439428.

Server Ops need to create the top-level repo for you, sdwilsh might be able to tell you how to import the whole history from cvs into hg if you want to have it.

Comment 9

[Serge Gautherie (:sgautherie)]

(True Hg repository with full history would be ideal.)

Comment 10

[Alex Vincent [:WeirdAl]]

Who wants to drive this?  I tried checking out Venkman (and only Venkman) code from CVS, and I wasn't able to.  I'm going to need this - or a source tarball I can get from FTP - fairly soon for Verbosio.

Comment 11

[Serge Gautherie (:sgautherie)]

(In reply to comment #10)
> Who wants to drive this?

I filed bug 478325 as the prerequisite.

> I tried checking out Venkman (and only Venkman) code from CVS, and I wasn't able to.

http://mxr.mozilla.org/comm-central/source/client.py?mark=12,271#268
does it for SeaMonkey.

sdwilsh seems to know how (and be able !?) to copy the needed files, with history.

Comment 12

[Serge Gautherie (:sgautherie)]

(Ah :-()

Comment 13

[Shawn Wilsher :sdwilsh]

I don't recall how I did it anymore, but there are tools to go from cvs->hg.

Comment 14

[:Gijs (he/him)]

(In reply to comment #10)
> Who wants to drive this?  I tried checking out Venkman (and only Venkman) code
> from CVS, and I wasn't able to.  I'm going to need this - or a source tarball I
> can get from FTP - fairly soon for Verbosio.

Huh? How does

cvs co mozilla/extensions/venkman

not work?

Comment 15

[:Gijs (he/him)]

Unless someone spots something criminally wrong with this repo:

http://hg.mozilla.org/users/gijskruitbosch_gmail.com/vnk-final/

I'll push that to the main hg.mozilla.org/venkman repo next week.

(yes, some of the timestamps for some checkins are weird. That's because the files were CVS 'moved', which means history was copied across with a new timestamp to reflect they did not exist in that location before that date. I'm not fixing this, it may be impossible (not sure if the original timestamps are somewhere to be found) and even if it is, it's not worth it, in particular because of the way HG works)

Comment 16

[Serge Gautherie (:sgautherie)]

(In reply to comment #9)
> (True Hg repository with full history would be ideal.)

(In reply to comment #15)
> http://hg.mozilla.org/users/gijskruitbosch_gmail.com/vnk-final/

At first glance, it looks great !

Comment 17

[Alex Vincent [:WeirdAl]]

I'd suggest adding a README file with instructions on what Venkman is, how to build Venkman, and how to package it as an XPI.  (I filed a bug on Inspector requesting the same a few weeks ago.)

We probably don't need any .cvsignore files anymore.  :-)

Comment 18

[:Gijs (he/him)]

(In reply to comment #17)
> I'd suggest adding a README file with instructions on what Venkman is, how to
> build Venkman, and how to package it as an XPI.  (I filed a bug on Inspector
> requesting the same a few weeks ago.)
> 
> We probably don't need any .cvsignore files anymore.  :-)

Right, but neither of these are issues with the import as such. These are all the files under CVS version control at the moment. Please file followup bugs for other issues. :-)

Comment 19

[:Gijs (he/him)]

~/devmoz/vnk/hgimport/finalpush/vnk-final gijskruitbosch$ hg push
pushing to ssh://hg.mozilla.org/venkman/
searching for changes
remote: adding changesets
remote: adding manifests
remote: adding file changes
remote: added 586 changesets with 1326 changes to 217 files

Will be filing a bunch of followup bugs now, apologies in advance for spam. :-)

Comment 20

[Serge Gautherie (:sgautherie)]

(In reply to comment #17)

Follow-up bugs filed.

Comment 21

[Serge Gautherie (:sgautherie)]

V.Fixed