On 03/17/2015 06:13 PM, Matěj Týč wrote: > Hello, > the subject says it all, m4 allows you to define and/or undefine macros, > so if one uses it e.g. in Makefiles, it is easy to alter the output. > > However, m4 is a real pain to use, m4sugar is so much better. But > autom4te, the convenient m4sugar "frontend", doesn't accept these > options, which I would like to change. I can elaborate more on use cases > if you are interested, but as this change is minor and > backwards-compatible, I hope that there will not be much resistance in > accepting it :-)
Seems reasonable to me, on that grounds. > > The patch is attached. > There are, however, some things I am not sure about - namely caching and > freezing. I don't have insight into them, but I have suspicion that they > may interfere with the new define/undefine stuff. Yeah, I'm also worried about that, and don't want to commit this code unless the testsuite is also enhanced to cover the cases. > > > autom4te.diff > While not mandatory, I find it nicer to send patches with 'git send-email', because it includes a diffstat summary of what files were touched in the patch (making it easier to see if a patch is trivial or potentially incomplete). In addition to my comment about needing testsuite additions, we also need documentation in the manual about the new command line options, and mention in NEWS about the new feature. And as this is a feature rather than a bug fix, it starts to be complicated enough that I'd feel better if you had FSF copyright assignment on file, or if someone else that already has copyright assignment can implement the feature from just your description. Shall we pursue that offline? So as not to taint myself (in case I'm the one to implement the feature), I have not read the rest of your patch email to see what you actually did. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
