module name is too long for sun-12.3 compiler

Bug #1658516 reported by Petr Koval
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Siesta
Triaged
Undecided
Unassigned

Bug Description

Dear devs,

I was trying to compile recent siesta with Sun 12.3 compiler (free of charge compiler suite) and it would not work because of some too long module name, see below

The arch.file is attached.

The version was 12.3, 12.4 is newer, but does not work quite well (for me).

Best regards,

Peter

kovalp:siesta.testing$ cd siesta/arch.make.sun-12.4.seq/
kovalp:arch.make.sun-12.4.seq$ make

Compilation architecture to be used: unknown
If this is not what you want, create the right
arch.make file using the models in Src/Sys

Hit ^C to abort...

==> Incorporating information about present compilation (compiler and flags)
make "FPPFLAGS=-DFC_HAVE_ABORT" compinfo.o
make[1]: Entering directory `/home/kovalp/programs/siesta.testing/siesta/arch.make.sun-12.4.seq'
sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT compinfo.F90
make[1]: Leaving directory `/home/kovalp/programs/siesta.testing/siesta/arch.make.sun-12.4.seq'

sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT /home/kovalp/programs/siesta.testing/siesta/Src/class_Fstack_Pair_Geometry_SpData2D.F90

module class_Fstack_Pair_Geometry_dSpData2D
       ^
"/home/kovalp/programs/siesta.testing/siesta/Src/class_Fstack_Pair_Geometry_SpData2D.F90", Line = 1, Column = 8: ERROR: The compiler has detected errors in module "CLASS_FSTACK_PAIR_GEOMETRY_DSPD". No module information file will be created for this module.
       ^
"/home/kovalp/programs/siesta.testing/siesta/Src/class_Fstack_Pair_Geometry_SpData2D.F90", Line = 1, Column = 8: ERROR: Module name length exceeds the maximum of 31 characters.

end module class_Fstack_Pair_Geometry_dSpData2D
           ^
"/home/kovalp/programs/siesta.testing/siesta/Src/class_Fstack_Pair_Geometry_SpData2D.F90", Line = 12, Column = 12: ERROR: Module name length exceeds the maximum of 31 characters.

f90comp: 543 SOURCE LINES
f90comp: 3 ERRORS, 0 WARNINGS, 0 OTHER MESSAGES, 0 ANSI
make: *** [class_Fstack_Pair_Geometry_SpData2D.o] Error 1

Tags: compiler
Revision history for this message
Petr Koval (kovalp) wrote :
Revision history for this message
Nick Papior (nickpapior) wrote :

Thanks!

Could you try with the f95 compiler from the sun suite (instead of f90)? It seems the name length limitations are released in that version (according to the standards):
https://docs.oracle.com/cd/E19205-01/819-5263/aevop/index.html

Revision history for this message
Petr Koval (kovalp) wrote : Re: [Bug 1658516] Re: module name is too long for sun-12.3 compiler

hm. I used sunf95 instead of sunf90 with the same error. I can promise to
try 12.4 at some point. It may seem as an unnecessary work (because the
program can be compiled with ifort and gfortran), but compiling with
different compilers/running is normally helping to elevate the quality of
software.

Best regards,

Peter

On Wed, Jan 25, 2017 at 7:41 AM, Nick Papior <email address hidden>
wrote:

> Thanks!
>
> Could you try with the f95 compiler from the sun suite (instead of f90)?
> It seems the name length limitations are released in that version
> (according to the standards):
> https://docs.oracle.com/cd/E19205-01/819-5263/aevop/index.html
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1658516
>
> Title:
> module name is too long for sun-12.3 compiler
>
> Status in Siesta:
> New
>
> Bug description:
> Dear devs,
>
> I was trying to compile recent siesta with Sun 12.3 compiler (free of
> charge compiler suite) and it would not work because of some too long
> module name, see below
>
> The arch.file is attached.
>
> The version was 12.3, 12.4 is newer, but does not work quite well (for
> me).
>
> Best regards,
>
> Peter
>
> kovalp:siesta.testing$ cd siesta/arch.make.sun-12.4.seq/
> kovalp:arch.make.sun-12.4.seq$ make
>
> Compilation architecture to be used: unknown
> If this is not what you want, create the right
> arch.make file using the models in Src/Sys
>
> Hit ^C to abort...
>
> ==> Incorporating information about present compilation (compiler and
> flags)
> make "FPPFLAGS=-DFC_HAVE_ABORT" compinfo.o
> make[1]: Entering directory `/home/kovalp/programs/siesta.
> testing/siesta/arch.make.sun-12.4.seq'
> sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT compinfo.F90
> make[1]: Leaving directory `/home/kovalp/programs/siesta.
> testing/siesta/arch.make.sun-12.4.seq'
>
> sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT
> /home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90
>
> module class_Fstack_Pair_Geometry_dSpData2D
> ^
> "/home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90", Line = 1, Column = 8: ERROR: The
> compiler has detected errors in module "CLASS_FSTACK_PAIR_GEOMETRY_DSPD".
> No module information file will be created for this module.
> ^
> "/home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90", Line = 1, Column = 8: ERROR: Module
> name length exceeds the maximum of 31 characters.
>
> end module class_Fstack_Pair_Geometry_dSpData2D
> ^
> "/home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90", Line = 12, Column = 12: ERROR: Module
> name length exceeds the maximum of 31 characters.
>
> f90comp: 543 SOURCE LINES
> f90comp: 3 ERRORS, 0 WARNINGS, 0 OTHER MESSAGES, 0 ANSI
> make: *** [class_Fstack_Pair_Geometry_SpData2D.o] Error 1
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/siesta/+bug/1658516/+subscriptions
>

Revision history for this message
Nick Papior (nickpapior) wrote :

I agree.
It would be best to support the sunf90 compiler as well. I am guessing there will be more problems.
I will try to fix these problems in the coming future. I am not currently easily accessible to different compilers.

Thanks for reporting.

Revision history for this message
Petr Koval (kovalp) wrote :
Download full text (3.8 KiB)

Hello Nick,

I succeed to compile with g95, and PGI went farther. PGI complains about
dsqrt() as you can see below.

pgf90 -c -O0 -DSIESTA__NO_MRRR -g
/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90
PGF90-S-0155-Intrinsic not supported in initialization: dsqrt
(/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90:
125)
PGF90-S-0155-Intrinsic not supported in initialization: dsqrt
(/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90:
126)
PGF90-S-0155-Intrinsic not supported in initialization: dsqrt
(/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90:
127)
PGF90-S-0155-Intrinsic not supported in initialization: dsqrt
(/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90:
128)
PGF90-S-0155-Intrinsic not supported in initialization: dsqrt
(/home/kovalp/programs/siesta.testing/siesta/Src/m_trialorbitalclass.f90:
134)

sunf90/sunf95 12.3 still complains about too long module names.

Best regards,

Peter

On Thu, Jan 26, 2017 at 8:02 AM, Nick Papior <email address hidden>
wrote:

> I agree.
> It would be best to support the sunf90 compiler as well. I am guessing
> there will be more problems.
> I will try to fix these problems in the coming future. I am not currently
> easily accessible to different compilers.
>
> Thanks for reporting.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1658516
>
> Title:
> module name is too long for sun-12.3 compiler
>
> Status in Siesta:
> New
>
> Bug description:
> Dear devs,
>
> I was trying to compile recent siesta with Sun 12.3 compiler (free of
> charge compiler suite) and it would not work because of some too long
> module name, see below
>
> The arch.file is attached.
>
> The version was 12.3, 12.4 is newer, but does not work quite well (for
> me).
>
> Best regards,
>
> Peter
>
> kovalp:siesta.testing$ cd siesta/arch.make.sun-12.4.seq/
> kovalp:arch.make.sun-12.4.seq$ make
>
> Compilation architecture to be used: unknown
> If this is not what you want, create the right
> arch.make file using the models in Src/Sys
>
> Hit ^C to abort...
>
> ==> Incorporating information about present compilation (compiler and
> flags)
> make "FPPFLAGS=-DFC_HAVE_ABORT" compinfo.o
> make[1]: Entering directory `/home/kovalp/programs/siesta.
> testing/siesta/arch.make.sun-12.4.seq'
> sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT compinfo.F90
> make[1]: Leaving directory `/home/kovalp/programs/siesta.
> testing/siesta/arch.make.sun-12.4.seq'
>
> sunf90 -c -O0 -DSIESTA__NO_MRRR -g -DFC_HAVE_ABORT
> /home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90
>
> module class_Fstack_Pair_Geometry_dSpData2D
> ^
> "/home/kovalp/programs/siesta.testing/siesta/Src/class_
> Fstack_Pair_Geometry_SpData2D.F90", Line = 1, Column = 8: ERROR: The
> compiler has detected errors in module "CLASS_FSTACK_PAIR_GEOMETRY_DSPD".
> No module information file will be created for this module.
> ^
> "/home/kovalp/programs/siesta.testing/siesta/Src/cla...

Read more...

Revision history for this message
Nick Papior (nickpapior) wrote :

Yes, I bet there are plenty more...

I would suggest you to fix the bugs on a separate branch for the pgi compiler (only). Then request a merge-request.

Thanks.

Nick Papior (nickpapior)
Changed in siesta:
status: New → Triaged
tags: added: compiler
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.