From: Alexey Pakhunov (alexeypa_at_[hidden])
Date: 2005-08-31 02:56:30
Andrey Melnikov wrote:
> I tried to decompose your megapatch. Here are the subpatches I see:
1. the new msplatformsdk feature (without corresponding support in the
2. new cpu-arch feature (without support)
3. configurations class
4. modification of check-init-parameters rule
5. new configure rule
6. "configure all" syntax
7. "version registration" framework
8. VC8 vcvarsall.bat stuff to enable cross-platform and
non-cross-platform compilation for amd64/EM-64T and Itanium/IA-64
9. interactions between platformsdk module and toolsets
10. manual generation of environment setup script for
11. trivial change of toolset.flags into just flags
12. migration to the CONFIG_COMMAND framework
13. registry-based autodetection
14. a trick to export environment variables from cmd into sh
15. single vsvars.bat run for link and mt on VC8
> There are 14 distinct almost independent changes, and each of the
> changes should be grounded enough in order to be accepted.
I'll try to highlight dependencies:
6 - depends on 5;
7 - depends on 3 and 5;
8 - depends on 2 and 7;
9 - depends on 2, 4, 7 and 12;
10 - depends on 9 and 12;
12 - depends on 7 and 14;
13 - depends of 7;
15 - is a duplicate of 12.
All "Depends" above mean that either functionality cannot be implemented
or a patch (AKA .diff) cannot be prepared before the previous parts will
be submitted. It is far from "almost independent".
> I think we should discuss and submit to CVS each change separately.
Agree. How I see this process:
1. We start to review the changes in the order of the list above. I'll
post first 4 (it is easy to separate them out) as 4 separate .diff files.
2. After submitting first 4 patches I'll again prepare the big one and a
couple of next subpatches.
3. We'll repeat 2 until everything will be submitted.
PS: One more thing I wanted to add. I agree that the patch is big. But
at least the half of it can be reviewed even now without any problems
and submitted later part by part. So let's stop talking about the size
of the path and let's start to review it.
Best regards/Venlig hilsen,
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk