TONYC TPF Grant 7 report 16

[Hours]         [Activity]
2016/07/04      Monday
 2.22           #127834 (sec) comments, fix some issues
 0.53           #127834 (sec) customized updates, testing, comment with
                new patchsets
 1.57           more parallel gmake
=====
 4.32

2016/07/05      Tuesday
 1.98           more parallel gmake
 2.47           more parallel gmake
=====
 4.45

2016/07/06      Wednesday
 0.40           #24000 research and comment
 1.97           more parallel gmake, fix search order conflict, repeat
                testing
 2.45           more parallel gmake, polish, performance testing,
                optimization
=====
 4.82

2016/07/07      Thursday
 1.75           more parallel gmake, more optimization, re-work deps
                closer to previous
 1.77           more parallel gmake, post as ticket 128564
=====
 3.52

Which I calculate is 17.11 hours

Approximately three issues were reviewed or worked on.

I spent much of this week building on bulk88's gmake build
parallelization work.

Both the dmake and gmake builds build miniperl and perl itself in
parallel, but then build all the nonxs extensions in a sequence which
is parallel with building all XS extenions in sequence.

This is a large improvement over the purely sequential build done
previously (and still done by nmake), but it fairly sub-optimal.

To allow extensions to build in parallel I wrote a simple makefile
generator to generate exts.mk for gmake - a makefile for building
extensions.

This significantly improved build time, but still needs some work.

• TONYC TPF Grant 7 report 16 by Tony Cook