Re: [boost] Re: [multi_index] problem with gcc 3.4.2 for Compaq Tru64UNIX
----- Mensaje original ----- De: Markus Schöpflin <markus.schoepflin@comsoft.de> Fecha: Viernes, Octubre 8, 2004 5:04 pm Asunto: [boost] Re: [multi_index] problem with gcc 3.4.2 for Compaq Tru64 UNIX
Markus Schöpflin wrote:
Joaquín Mª López Muñoz wrote:
mips-tfile, /tmp//cctFjnCI.s:10965 string too big (35326 bytes) line:? #.stabs?"and_<boost::mpl::not_<boost::is_same<boost::mpl: [very long expression]
A quick google search indicates that this is a problem with gcc on alpha when compiling with debug symbols. A possible workaround might be to use -gstabs instead of -g, I'll try it ASAP.
I did some experiments and the only way to compile these files it to disable debug symbols completely. I have done that on the toolset for gcc used for the regression runs, I hope it helps to clean up some of the errors.
Hi Markus, I see that the failures are gone, thank you! Would you be so kind as to explain to me which compiler options you touched to make this work? I'd like to add that information on the "Compiler specifics" section of Boost.MultiIndex docs. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo
JOAQUIN LOPEZ MU?Z wrote:
Would you be so kind as to explain to me which compiler options you touched to make this work? I'd like to add that information on the "Compiler specifics" section of Boost.MultiIndex docs.
See the note at http://www.meta-comm.com/engineering/boost-regression/developer/summary.html for a short explanation. The exact compiler switch removed was "-g". I will try to find out whether there is perhaps an updated assembler available which fixes the problem. Markus
Markus Schöpflin ha escrito:
JOAQUIN LOPEZ MU?Z wrote:
Would you be so kind as to explain to me which compiler options you touched to make this work? I'd like to add that information on the "Compiler specifics" section of Boost.MultiIndex docs.
See the note at http://www.meta-comm.com/engineering/boost-regression/developer/summary.html for a short explanation.
Oh, thanks, forgot to check that out.
The exact compiler switch removed was "-g". I will try to find out whether there is perhaps an updated assembler available which fixes the problem.
Please keep me informed if you make any progress. Thank you again. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo
Joaquín Mª López Muñoz wrote:
Markus Schöpflin ha escrito:
The exact compiler switch removed was "-g". I will try to find out whether there is perhaps an updated assembler available which fixes the problem.
Please keep me informed if you make any progress. Thank you again.
More on this... It's really not the fault of the system assembler that the file cannot be assembled by gcc in debug mode but the fault of mips-tfile which is an executable (first needed for MIPS, now only in use for Tru64, hence the name) which comes with gcc and is used to put the debug symbols from the assembler source into the object file _after_ the system assembler has created the object file. The length of the debug symbols mips-tfile can handle is limited to 32768 bytes. :-( There are some defines in the source code of mips-tfile.c (namely PAGE_SIZE and PAGE_USIZE) which look like they could be used to change that limitation but I'm not sure what else depends on these constants. I thought you might be interested in these findings. Markus
participants (3)
-
JOAQUIN LOPEZ MU?Z -
Joaquín Mª López Muñoz
-
Markus Schöpflin