Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-12 Thread Sergio Durigan Junior
On Monday, July 07 2014, Matthias Klose wrote: My request, then, is that you make this package available to the architectures where GDB is available. on I hope you find this reasonable. thanks, talking with mjw I think this is reasonable. however for debian multiarch cross builds this

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-07 Thread Sergio Durigan Junior
Which architectures don't need it? Any architecture that support glibc and gdb for example benefits from having sdt markers available. Mark is right. It seems it is never enough to repeat this, but sys/sdt.h should not be related to SystemTap. GDB makes a strong use of this header file, and

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-07 Thread Matthias Klose
Am 07.07.2014 22:55, schrieb Sergio Durigan Junior: Which architectures don't need it? Any architecture that support glibc and gdb for example benefits from having sdt markers available. Mark is right. It seems it is never enough to repeat this, but sys/sdt.h should not be related to

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-05 Thread Matthias Klose
Control: severity -1 serious the severity of this issue was changed by a non-maintainer, without giving any reason. re-raising the severity of the issue, and preparing a NMU to move the header file to an architecture specific location. -- To UNSUBSCRIBE, email to

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-05 Thread Mark Wielaard
Hi Matthias, On Sat, 2014-07-05 at 17:01 +0200, Matthias Klose wrote: re-raising the severity of the issue, and preparing a NMU to move the header file to an architecture specific location. What is the issue you are seeing? I thought that what you saw was something unrelated to sys/sdt.h. And

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-05 Thread Matthias Klose
Am 05.07.2014 18:21, schrieb Mark Wielaard: Hi Matthias, On Sat, 2014-07-05 at 17:01 +0200, Matthias Klose wrote: re-raising the severity of the issue, and preparing a NMU to move the header file to an architecture specific location. What is the issue you are seeing? I thought that what

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-07-05 Thread Mark Wielaard
On Sat, 2014-07-05 at 18:32 +0200, Matthias Klose wrote: could you tell me why you need the header on architectures that don't need it? Which architectures don't need it? Any architecture that support glibc and gdb for example benefits from having sdt markers available. -- To UNSUBSCRIBE,

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-06-11 Thread Matthias Klose
Control: reopen -1 my first assumption turned out wrong as Samuel noted. However I still think it is wrong to place this file into /usr/include in an architecture independent package so that it is seen by every cross compiler. People did notice this earlier in

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-20 Thread Mark Wielaard
On Mon, 2014-05-19 at 22:53 +0200, Matthias Klose wrote: Am 19.05.2014 21:00, schrieb Mark Wielaard: It is just the package name that refers to systemtap, but it could as well have been called gdb-sdt-devel for example. In which case it should at least work as is on any arch gdb

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-20 Thread Matthias Klose
Am 20.05.2014 10:57, schrieb Mark Wielaard: I am just not clear what the precise bugs are that you are seeing. The gcc example is somewhat hard to understand. Is the issue you are seeing with gcc really caused by sys/sdt.h or might it be the g++ template decl ordering problem discussed here:

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-20 Thread Timo Juhani Lindfors
Hi, Matthias Klose d...@debian.org writes: I'm not complaing about the name of the package, but that it apparently *does* have some unintended effects on some architectures. is there something simpler than gcc that FTBFS? I'd like to look into the issue but gcc is quite heavy to build,

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-19 Thread Matthias Klose
Package: systemtap Version: 2.3-2 Severity: serious Tags: sid jessie The sys/sdt.h header file is shipped in an architecture independent package, and installed into /usr/include where it is found on the include path for every architecture. Seen that the mere inclusion of this header causes build

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-19 Thread Timo Juhani Lindfors
Hi, [ Adding reporter of #726248 to CC and quoting the bug report fully for him. ] Matthias Klose d...@debian.org writes: The sys/sdt.h header file is shipped in an architecture independent package, and installed into /usr/include where it is found on the include path for every

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-19 Thread Mark Wielaard
On Mon, 2014-05-19 at 20:17 +0200, Matthias Klose wrote: The sys/sdt.h header file is shipped in an architecture independent package, and installed into /usr/include where it is found on the include path for every architecture. [...] what about issues on architectures not supported by

Bug#748694: sys/sdt.h is architecture specific, and causing issues on unsupported architectures

2014-05-19 Thread Matthias Klose
Am 19.05.2014 21:00, schrieb Mark Wielaard: On Mon, 2014-05-19 at 20:17 +0200, Matthias Klose wrote: The sys/sdt.h header file is shipped in an architecture independent package, and installed into /usr/include where it is found on the include path for every architecture. [...] what about