Discussion:
[XOM-interest] java.net migration
Elliotte Rusty Harold
2011-01-22 22:25:45 UTC
Permalink
java.net is locked and read-only for XOM, and just when I was ready to
spend some time fixing some issues, and adding the biggest new feature
in some time (Canonical XML 1.1 support).

I have received no estimates from the java.net administrators as to
when the new site will be ready for business.

I wonder how long it would take to import everything into Google code?
--
Elliotte Rusty Harold
elharo at ibiblio.org
Subhash Chandran
2011-01-23 04:40:58 UTC
Permalink
If you are planning Mercurial as source repo, it is as simple as:

1. Configuring $HOME/.hgrc: Add following:

[extensions]
hgext.convert=

2. Converting the repo:

$ hg convert cvs-working-copy/ local-hg-repo/

3. Pushing the repo to the Google code.

I tried converting the latest XOM cvs repo and it works like magic. More
details on 'hg convert' can be found here:

http://mercurial.selenic.com/wiki/ConvertExtension

Subhash.

On Sun, Jan 23, 2011 at 3:55 AM, Elliotte Rusty Harold
Post by Elliotte Rusty Harold
java.net is locked and read-only for XOM, and just when I was ready to
spend some time fixing some issues, and adding the biggest new feature
in some time (Canonical XML 1.1 support).
I have received no estimates from the java.net administrators as to
when the new site will be ready for business.
I wonder how long it would take to import everything into Google code?
--
Elliotte Rusty Harold
elharo at ibiblio.org
_______________________________________________
XOM-interest mailing list
XOM-interest at lists.ibiblio.org
http://lists.ibiblio.org/mailman/listinfo/xom-interest
--
Regards,
Subhash Chandran S

Founder http://FileIt.in/
Elliotte Rusty Harold
2011-01-23 14:37:46 UTC
Permalink
I'm not planning to use mercurial. SVN seems fine for XOM's needs, and
I really don't want to learn a sixth (or is that seventh) source code
control system. :-)
--
Elliotte Rusty Harold
elharo at ibiblio.org
c0d3g33k
2011-01-24 14:00:05 UTC
Permalink
I'm not planning to use mercurial. SVN seems fine for XOM's needs, and I really don't want to
learn a sixth (or is that seventh) source code control system. :-)
You should at least consider it since the opportunity coincides with the already planned migration
and, as was demonstrated, the switch is trivial. The Mercurial user interface is similar enough to
SVN that you can probably carry on much as you have without much additional learning or change in
personal workflow.

The thing is, it's not only XOM's (ie. ERH's) needs that a move to Mercurial would affect. It may
very well improve the experience of members of the XOM community, and for a FLOSS project that
should matter.

I should note this isn't about blind advocacy of the tool du jour. For the typical FLOSS use case
(track upstream while maintaining local changes that aren't ready or suitable for upstream), using
one of the recent distributed version control systems has improved the experience quite a lot
compared to CVS/SVN. Not in a revolutionary way, but it removes some of the 'friction' from the
process, and that makes a bigger difference than one might expect (than I expected, at least).
Michael Kay
2011-01-24 14:23:00 UTC
Permalink
Post by c0d3g33k
You should at least consider it since the opportunity coincides with the already planned migration
and, as was demonstrated, the switch is trivial. The Mercurial user interface is similar enough to
SVN that you can probably carry on much as you have without much additional learning or change in
personal workflow.
My experience was that switching from Subversion to Mercurial involved a
substantial learning curve and also a change in development processes -
you have to work the way the tool expects you to work. In fact, after a
week we gave up - it may well have been worth persevering, but we ran
out of patience. The Mercurial documentation does warn you that this
experience is not unusual.

Michael Kay
Saxonica
c0d3g33k
2011-01-24 16:37:08 UTC
Permalink
My experience was that switching from Subversion to Mercurial involved a substantial learning
curve and also a change in development processes - you have to work the way the tool expects you
to work. In fact, after a week we gave up - it may well have been worth persevering, but we ran
out of patience. The Mercurial documentation does warn you that this experience is not unusual.
Yeah, I would expect that to be the case for a team. I would note that it isn't so much that the
tool imposes changes on a development team, it's that this class of tool was created to enable a
different way of working. That's naturally going to mean changes if the tool is to be used fully
and not just as a better drop-in for the old tool (which pretty much describes SVN).

My original comment was based on the fact that Mercurial is intentionally similar to SVN in command
syntax, particularly for the most common commands. For the simplest case of a single master
repository with a single (or very few) committer, the learning curve is pretty shallow. Not flat,
mind you, but shallow. When giving it a try for personal projects based on SVN, Mercurial was easy
to transition to and much less painful than Git, though they are quite similar in general philosophy
(though Git is nice to use as well once the command syntax is grokked, and at some level
interchangable with Mercurial). The point was that for XOM, which has a fairly standard and simple
repository structure and (as far as I can tell from a quick glance at the online repo) a single
committer without a complex team to deal with, the transition would perhaps not be so painful.

I suppose there's little point to switching anyway. To be honest, DVCS is a tool of the bazaar, and
XOM has more of a cathedral-like air about it.

So never mind - stick with SVN.

Loading...