[PATCH stable] setup: set SystemRoot even if .hg doesn't exist (issue3073)

[Mads Kiilerich]

Yuya Nishihara wrote, On 10/30/2011 02:48 PM:
> Mads Kiilerich wrote:
>> Yuya Nishihara wrote, On 10/30/2011 02:18 PM:
>>> # HG changeset patch
>>> # User Yuya Nishihara<yuya at tcha.org>
>>> # Date 1319980502 -32400
>>> # Branch stable
>>> # Node ID 4a702fd4fcf5f161af2ea49d543eb661add84a6f
>>> # Parent  bf55991af17e7e07593cceb2163328ae8cf5ee65
>>> setup: set SystemRoot even if .hg doesn't exist (issue3073)
>> Is there any reason we don't move the whole env initialisation outside
>> the .hg check? Do buildhgextindex have different needs than other
>> invocations of hg?
> There's no particular reason to move *only* SystemRoot, but that's just
> because buildhgextindex doesn't need the other environment variables
> than SystemRoot. It doesn't run hg command itself.

buildhgextindex will invoke python and use the Mercurial modules. That 
is not very different from running the hg command.

Pure _is_ apparently needed for running setup.py build_hgextindex.

And why is LD_LIBRARY_PATH not needed for buildhgextindex if it is 
needed for running hg?

/Mads