I think that disabling the library build of a given library should certainly disable its tests. However, I still don't understand why someone would want to disable a generator in the compiler. What does it cost you to have the generator as an option?
On Wed, Aug 18, 2010 at 11:40 PM, Anthony Molinaro < antho...@alumni.caltech.edu> wrote: > I'm of the opinion that it should be possible to disable a generator, but > I think it should be the same switch as disables the runtime library for > a particular language. I think this makes the most sense, since I don't > see many reasonse to generate bindings which you can't actually compile > and run. I'll poke around through the build system to check, but I'm > pretty confident it should be very possible to use the same configure > option to disable the generator, the building of the runtime library > and the running of the tests. > > I think 0.4 can probably be released without this, but I think it's a > good goal for 0.5. > > -Anthony > > On Thu, Aug 19, 2010 at 07:36:03AM +0200, Bruce Lowekamp wrote: > > I'm ok with not blocking the release on it. Maybe there should be release > notes (don't disable generators, don't run bootstrap). I think I would have > run into it because I think the first time I built thrift I had to disable > something to get it to work, but if the goal is for faster releases, I think > I'm ok with that. > > > > Bruce > > > > Sent from my iPhone > > > > On Aug 18, 2010, at 21:55, Bryan Duxbury <br...@rapleaf.com> wrote: > > > > > Bruce - I commented on your ticket, and I agree that we should make > some > > > changes. However, do you really think this is important enough for us > to > > > torpedo the RC? I don't see this as something critical enough that we > would > > > have made it a blocker for the release. > > > > > > On Wed, Aug 18, 2010 at 4:24 PM, Bruce Lowekamp < > bruce.lowek...@skype.net>wrote: > > > > > >> -1 > > >> > > >> Sorry to pop in so late. I've been meaning to get to tracking thrift > > >> development, and the last couple days seemed like a good opportunity, > so > > >> I've been catching up on it and merging some of our stuff. > > >> > > >> We build with quite a few of the generators disabled. The patch for > > >> THRIFT-133 to parse/t_program.h (r980498) added a check if the > generator in > > >> a namespace declaration is known and throws if not. Makes sense, but > this > > >> makes it impossible to pass the sanity checks that are run by "make > install" > > >> in test when generators are disabled (ThriftTest.thrift requires java, > cpp, > > >> rb, perl, csharp, js). > > >> > > >> Obviously re-enabling the generators is pretty easy, but seems like > either > > >> the configure flags for generators should be removed, the tests should > be > > >> modified so that the namespace declarations are auto-generated based > on > > >> configuration, or warnings should be added to the configure > > >> > > >> I'd personally be in favor of removing the --disable-gen flags for > "core" > > >> libraries. (that are specified in the test.thrift files). Though > > >> pre-processing them isn't that bad of an idea. > > >> > > >> Somewhat minor/corner-case issue. Seems like it might be a little > > >> unfriendly to newcomers who experiment, though. > > >> > > >> I'll open a jira issue. > > >> > > >> Bruce > > >> > > >> > > >> > > >> On Aug 18, 2010, at 1:59 PM, ro...@bufferoverflow.ch wrote: > > >> > > >>> +1 > > >>> > > >>> I've built the following languages successfully: > > >>> cpp, java, erlang, ruby, perl, python > > >>> > > >>> and the following available Unit Tests passed: > > >>> cpp, java > > >>> > > >>> see: > > >>> > > >> > http://www.bufferoverflow.ch/hudson/job/Thrift-0.4.0/default/lastStableBuild/ > > >>> > > >>> > > >>> > > >>> > > >>> Quoting Bryan Duxbury <br...@rapleaf.com>: > > >>> > > >>>> Hi all, > > >>>> > > >>>> As discussed, in all the time it took for us to finally release > Thrift > > >> 0.3, > > >>>> we've accumulated enough changes for Thrift 0.4! > > >>>> > > >>>> I propose we accept > > >>>> http://people.apache.org/~bryanduxbury/thrift-0.4.0-rc1.tar.gz<http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz> > <http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz> > > >> <http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz>as > > >>>> the official Thrift 0.4.0 release. > > >>>> > > >>>> I created this tarball by checking out URL: > > >>>> https://svn.apache.org/repos/asf/incubator/thrift/branches/0.4.xand > > >> running > > >>>> "./bootstrap.sh && ./configure && make dist". > > >>>> > > >>>> The GPG signature for this tarball can be found at > > >>>> http://people.apache.org/~bryanduxbury/thrift-0.4.0-rc1.tar.gz.asc<http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz.asc> > <http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz.asc> > > >> <http://people.apache.org/%7Ebryanduxbury/thrift-0.4.0-rc1.tar.gz.asc > >. > > >>>> It has an MD5 signature of bb8466e5b884e9edef862a64f57abe38. > > >>>> dc. > > >>>> > > >>>> Please download, verify the signatures, untar, and install and test > the > > >>>> libraries you are familiar with. We need your +1 votes to make the > > >> release! > > >>>> If you encounter any problems, please let me know immediately. This > vote > > >>>> will close 72 hours from now. > > >>>> > > >>>> This release consists of 45 tickets that cover many bugs and > performance > > >>>> improvements. Please refer to the CHANGES file for the full list. In > > >>>> addition, I've rolled back a commit to the PHP extension that was > > >> causing > > >>>> build errors on non-Mac platforms. > > >>>> > > >>>> -Bryan > > >>>> > > >>> > > >>> > > >>> > > >>> ---------------------------------------------------------------- > > >>> This message was sent using IMP, the Internet Messaging Program. > > >>> > > >> > > >> > > > > -- > ------------------------------------------------------------------------ > Anthony Molinaro <antho...@alumni.caltech.edu> >
