Bug#758184: boinc: FTBFS due to dependency errors in Makefile

Gianfranco Costamagna costamagnagianfranco at yahoo.it
Tue Aug 26 07:03:01 UTC 2014


(please excuse my top posting)

Hi Guo and Steffen, 
sorry if it is taking too long, but you don't know the kind of problems I spotted in this deep code review! the makefiles are now consistent, or at least they should be. The problem was that we (I, because the gcc-4.9 patch was from me) patched boinc always in the wrong way, we had problems, differences between static and dynamic builds and we solved by removing flags.

this isn't the right way, this should be handled automatically by libtool. In fact now with the fixed makefiles and no debian patches I have able to parallel build it (make -j 8).

I spotted three bad debian patches, and one bad upstream file.

md5.c is not getting exported correctly (this is why probably we disabled fcgi, it wasn't able to find md5_init and so on). for some reason g++ when compiling a file .c behaves differently as it does when compiling the same cpp file. This lead to important functions not build, or not linked against boinc code, and hidden by our dynamic builds. I'm providing a patch for upstream today (as soon as I reach a computer to test if really renaming a file makes g++ stop exporting __cplusplus define in md5.h)

After this I think my review should be done, I don't want to push anything on unstable because it will require a seti rebuild, so please wait some days more :-)

We already have something in testing, so I don't think we should hurry up (freeze is approaching I know)

So in the next few hours/days I'll send something to you and hopefully upload :-)

stay tuned!

Gianfranco

Sent from Yahoo Mail on Android

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-boinc-devel/attachments/20140826/c8d2d607/attachment-0003.html>


More information about the pkg-boinc-devel mailing list