x64 hg.exe

[Mads Kiilerich]

On 22/06/12 13:15, Adrian Buehlmann wrote:
> On 2012-06-22 13:08, Mads Kiilerich wrote:
>> On 22/06/12 11:30, Adrian Buehlmann wrote:
>>> On 2012-06-22 03:42, Adrian Buehlmann wrote:
>>>> Copy the hg file to hg.py, replace the first line with
>>>>
>>>>     #!C:\Windows\system32\python27.dll
>>>>
>>>> (for a global installed Python 2.7)
>>>>
>>> The hg file (mercurials python script in the repo root) needs to be
>>> *renamed* to hg.py (not copied), otherwise the suite (MSYS shell) will
>>> prefer the hg over the hg.exe for most of the calls in the tests, which
>>> works because of the
>>>
>>>     #!/usr/bin/env python
>>>
>>> on the first line of the hg file (after all, python is in on PATH and
>>> there is a env.exe in C:\MinGW\msys\1.0\bin, which is on PATH as well).
>> Yes, we have discussed that before. The wiki used to say that 'hg'
>> should be removed and replaced by 'hg.exe' and 'hg.py', but I removed
>> that prerequisite in
>> http://mercurial.selenic.com/wiki/HackableMercurial?action=diff&rev1=29&rev2=30
>> .
> [..]
>
>> I changed the description partly to be compatible with the setup you
>> apparently are using, and partly because
>> http://selenic.com/repo/hg/rev/30cfa286947c implicitly introduce a
>> requirement to having 'hg'.
> Can you explain to me why 30cfa286947c should have introduced a
> requirement to have a hg?
>
> I don't see it.

Removing the hghave enabled the existing line
   $ PATH="$BINDIR" $PYTHON "$BINDIR"/hg merge -r 2

That hack has a bit of history and cause quite a bit of complexity in 
run-tests. It could probably be replaced with a more elegant hack now.

/Mads