mercurial at 10000: 6 new changesets (2 stable)

[Steve Borho]

On Thu, Dec 10, 2009 at 9:17 AM, Martin Geisler <mg at lazybytes.net> wrote:
> Dirkjan Ochtman <dirkjan at ochtman.nl> writes:
>
>> On Thu, Dec 3, 2009 at 22:15, Mercurial Commits <hg at intevation.org> wrote:
>>> http://selenic.com/repo/hg//rev/f91e5630ce7e
>>> changeset:   9999:f91e5630ce7e
>>> user:        Dan Villiom Podlaski Christiansen <danchr at gmail.com>
>>> date:        Tue Dec 01 16:06:10 2009 +0100
>>> summary:     setup: install translation files as package data
>>
>> I'm not really seeing what this buys us for the templates dir, other
>> than massive confusion for everyone who has done anything custom with
>> templates. Is this really necessary?
>
> The templates are installed in the same location as always, that is,
> inside the mercurial/ directory.
>
> What this changes is the setup.py code: before this, setup.py had a hack
> to work around the lack of package_data in Python 2.3. The hack simply
> turned data_files into package_data which means that everything was
> installed relative to mercurial/.
>
> After "de-hacking" this, we can now again use the data_files keyword
> argument to install files outside of mercurial/.
>
> Well, that's the idea at least... we've since then seen that it might
> not be a good idea to use data_files for the zsh and bash completion
> scripts (since we don't know where they should go on all platforms).

If I run 'hg st' after 'python setup.py py2exe' on Windows, it shows a
whole bunch of missing (!) files.  Is this expected nowadays?   The
tip of hg-main is really borked on Windows, I have to re-clone the hg
repository every time I build it.

--
Steve Borho