Hi Ralf. As you asked, I report here what would be, in my opinion,
the preferred order of review for the still-pending patches of mine.
I also add some details on the patches, where I think that might help.
1. [PATCH] "Work around a bug in Solaris make's file-inclusion
mechanism."
Simple fix for a simple bug (bug found by me, fix found by you,
final patch by me).
There is still an issue regarding whether we should squash even
multiple leading slashes in names of dependency files included
by make. I think we shouldn't, and I explained why (see last
mail in thread). Mostly a bike-shedding-ish issue, though.
2. [PATCH] "Overhauled and modularized tests in `instspc.test'"
Break up ugly and fragile test script "instspc.test" into multiple
(90!) autogenerated test scripts. Code shared by new tests placed
in new file `tests/instspc.sh'. Some hacks required to work around
make limitations (hit in auto-generation of tests).
Lately rebased against master (see last mail in the thread).
Tested quite extensively. No review done yet, not even partial.
3. [PATCH] "New tests on obsoleted usages of automake/autoconf
macros."
Already reviewed, minor details to be decided (whether autotools
caching is relevant for test `backcompat2.test', and whether test
`backcompat6.test' is too ugly/paranoid to be kept). Might be
placed in a new branch (I'd like to add new back-compatibility
tests in the future); not a very important issue ATM though.
4. [PATCH 0/3] "Some improvments to cscope support and testing"
Most issues with this patch series had already been solved. Might
require a rebase though, and a decision to which branch it should
be applied (master or dr-cscope?). Not very important, but it's
starting to get old, and I'd prefer to avoid useless bitrotting.
5. [PATCH + QUESTION] "Improve and extend tests `man*.test'."
Not reviewed yet. Among other things, it offers a better
exposition of a bug in BSD make. The poorly choiced title make
it easy for reviwers to miss the thread, so better not let it
to fade off even more.
6. [PATCH 0/6] Add support for non-default autotools in rebuild rules.
A real fix for a quite subtle bug in rebuild rules. This bug would
only affect developers with somewhat unusual (but legitimate) setup.
It's affecting me, in fact ;-)
You spotted some possibly serious problems in the original patch
series (test scripts hanging!). I fixed those and did more testing,
so there shouldn't be any serious bug anymore (I hope). It will
require a quite thorough review though, as the isue fixed by the
series is IMHO quite subtle.
That's all for the moment, I guess. You might prioritize the other
pending patches as you prefer, or we might come up later with a
follow-up prioritization of still-pending patch. No point in thinking
about that ATM.
Thanks,
Stefano
