[Build-common-hackers] a WAF class for CDBS
Rémi Thebault
remi.thebault at gmail.com
Thu Dec 30 18:13:19 UTC 2010
Hi Jonas
> A few comments on your recent changes:
>
>
> Arrrgh! Please be careful to commit each single logical code change
> separately, with a concise commit message - so we are best able to see
> via git changelog what has actually happened with the code. Better with
> one commit too many than one too few! You (again!) committed
> autogenerated autotools files together with hand-coded changes, and
> changes with more invasive changes than the corresponding commit
> messages indicate!
Oups Sorry I just thought this was one logical change. I mean adding a
script AND including it in Makefile.am.
I 'll be more cautious in future
> Please avoid ifdef contruct! I wrote bfore about this: it should
> generally be avoided as it causes order of snippet inclusion to matter,
> which is bad. If you do not understand what I mean, then say so, and I
> can try clarify. If you disagree then say so, and we can discuss it.
> Just being silent about my commenting on this while ignoring it, is
> annoying!
Oups sorry again! I don't disagree and do understand. I just did it quite fast and changed it in last commit
> >I just have a little concern : the messages are drawn in too many
> >output since I do the tests with shell if and everything is echoed.
> >I could put all those test and unpacking in a separate shell script and
> >echo only what I need. Or do you prefer implementation only in
> >makefile?
>
> I prefer using make $(warning ) and $(error ).
> Alternatively you can see utils.mk for an example of how to emit
> pseudo-code while silencing more convoluted actual code including
> shell-based error message. But I don't recommend that - better to
> simplify by using make instead of shell.
I did much better. No pollution output with only make instructions and
target-time var expansion .
I had to make some odd constructions to test shell commands return code
(like $(shell if cmd > /dev/null 2>&1; then echo something; fi)) but its
working fine.
>
> Did you look into the relevancy of doing checksum of individual files
> inside the wad instead?
>
I will soon !
>
> - Jonas
>
> _______________________________________________
> Build-common-hackers mailing list
> Build-common-hackers at lists.alioth.debian.org
> http://lists.alioth.debian.org/mailman/listinfo/build-common-hackers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/build-common-hackers/attachments/20101230/228bab0a/attachment.htm>
More information about the Build-common-hackers
mailing list