Bugzilla Bug 135227

Scilab 4.0 fails to build w/ gcc 4.1 complaining there's no fortran compiler on
the system:

checking for g77... no
checking for f2c... no
configure: error: Unable to configure: no Fortran compiler found

!!! Please attach the following file when filing a report to bugs.gentoo.org:
!!! /var/tmp/portage/scilab-4.0/work/scilab-4.0/config.log

!!! ERROR: sci-mathematics/scilab-4.0 failed.
Call stack:
  ebuild.sh, line 1531:   Called dyn_compile
  ebuild.sh, line 931:   Called src_compile
  scilab-4.0.ebuild, line 79:   Called econf '--without-tk' '--without-xaw3d'
'--with-gtk2' '--without-ocaml' '--without-java'
'--with-atlas-library=/usr/lib64' '--with-gfortran'
  ebuild.sh, line 533:   Called die

I can attest I have gfortran installed:

omegaone gralves # gfortran
gfortran: no input files

I'll attach config.log to this bug.

Created an attachment (id=88155) [edit]
Config Log

Thanks for the fast reply and I'll have a look at this tomorrow. Probably
a problem for x86_64. Interestingly, according to your config.log gfortran
is detected

configure:2037: checking for gfortran
configure:2053: found /usr/x86_64-pc-linux-gnu/gcc-bin/4.1.1/gfortran
configure:2064: result: yes

but somehow not used later on.

Talk to you soon,
Markus

Looks like there's some empty variable here:

configure:4968: checking for Fortran 77 compiler version
configure:4971:  --version </dev/null >&5
./configure: line 4972: --version: command not found
configure:4974: $? = 127
configure:4976:  -v </dev/null >&5
./configure: line 4977: -v: command not found
configure:4979: $? = 127
configure:4981:  -V </dev/null >&5
./configure: line 4982: -V: command not found
configure:4984: $? = 127
configure:4991: checking whether we are using the GNU Fortran 77 compiler
configure:5005:  -c  conftest.F >&5
./configure: line 5006: -c: command not found

This problem also occurs for me, any idea how to fix it?

Please note that scilab-2.7-r3 doesn't compile either.

These are the interesting (I think :-) output lines of "emerge scilab":

which: no g77 in
(/usr/local/sbin:/sbin:/usr/sbin:/usr/lib/portage/bin:/usr/local/bin:/bin:/usr/bin:/opt/bin:/usr/i686-pc-linux-gnu/gcc-bin/4.1.1:/opt/blackdown-jdk-1.4.2.03/bin:/opt/blackdown-jdk-1.4.2.03/jre/bin:/usr/kde/3.5/sbin:/usr/kde/3.5/bin:/usr/qt/3/bin:/usr/kde/3.4/sbin:/usr/kde/3.4/bin)
 * No fortran compiler found on the system!
 * Please add fortran to your USE flags and reemerge gcc!

!!! ERROR: sci-mathematics/scilab-2.7-r3 failed.
Call stack:
  ebuild.sh, line 1562:   Called dyn_setup
  ebuild.sh, line 665:   Called pkg_setup
  scilab-2.7-r3.ebuild, line 41:   Called die

!!! (no error message)
!!! If you need support, post the topmost build error, and the call stack if
relevant.

Linking g77 to gfortran-4.1.1 in /usr/bin works but its a horrible fix. Hope
there is a better one soon. I am personally having a problem with Xaw3d not
being found correctly. I am certain I have it installed but the configure
script is having trouble making it work.

(In reply to comment #6)
> Linking g77 to gfortran-4.1.1 in /usr/bin works but its a horrible fix. Hope
> there is a better one soon. I am personally having a problem with Xaw3d not
> being found correctly. I am certain I have it installed but the configure
> script is having trouble making it work.
> 

Could you please clarify if
1) You have /usr/bin/gfortran but need /usr/bin/g77 linked
to gfortran to compile it, or 
2) there is no /usr/bin/gfortran whatsoever on your system.

Regarding Xaw3d: Could you make sure that you have libXaw3d.so
on your system and please post the relevant part of your config.log.
Also, please post your emerge --info.

Thanks,
Markus

Upstream seems to have fixed the bug at some point, since a snapshot does
compile. I have other issues with those, but they do compile. The makefile
differences are too big for me to see what is needed.

(In reply to comment #8)
> Upstream seems to have fixed the bug at some point, since a snapshot does
> compile. I have other issues with those, but they do compile. The makefile
> differences are too big for me to see what is needed.
> 

Hi Maarten,

Thanks, I'll have a look at this over the weekend. However, since I don't
have this problem it might be difficult to track down. Could you please
post your emerge --info and the config.logs of the broken and working
scilab versions.

Thanks in advance,
Markus

Created an attachment (id=97089) [edit]
scilab-4.0 config.log

This one fails.

Created an attachment (id=97090) [edit]
scilab snapshot(20060913) config log 

This one does detect the fortran compiler.

Config logs added as requested.

Created an attachment (id=97153) [edit]
potential gfortran fix for x86_64

Thank you very much for the config files. I've come up with a
patch that might fix your problems. Unfortunately, I can't test
it lacking an x86_64 box. Please give it a shot and report back.

Thanks,
Markus

Created an attachment (id=97164) [edit]
Ebuild diff needed to get it to compile.

The patch works, but some additional things were needed.

(In reply to comment #14)
> Created an attachment (id=97164) [edit]
> Ebuild diff needed to get it to compile.
> 
> The patch works, but some additional things were needed.
> 

Ahh, I missed the Makefile.linux64. However, the $(get_libdir)
is the proper way to do it. If you need that mylibdir hack as in
your ebuild there's likely a problem with your install. E.g. 
atlas installs into /usr/$(get_libdir); if yours is somewhere
else you should check into this. 
Please let me know how this goes.

Best,
Markus

Created an attachment (id=97167) [edit]
Updated patch to fix Makefile.linux64 issue

Atlas is fine, it's just that scilab tries to install into /usr/lib and then
all the sed'ing happens in (image) /usr/lib64 because of my libdir. Guess i
didn't need to the atlas, but i just did all of them.

I meant: Guess i didn't have to change all of them in an attempt to get it to
compile.

So it's a scilab problem, rather than a atlas problem.

(In reply to comment #18)
> I meant: Guess i didn't have to change all of them in an attempt to get it to
> compile.
> 
> So it's a scilab problem, rather than a atlas problem.
> 

Ah, I understand now: scilab installs into /usr/lib whereas
it should go into /usr/lib64. I'll have to have a look
at the makefile to see if I can fix this.

Thanks for your help!

Created an attachment (id=97183) [edit]
patch to force proper multilib install patch

Could you please check if the following patch to the ebuild
helps.

Thanks!

Patches work.

Out of curiosity, do you get (if using the tcltk interface):

Scilab : X error trapped - error message follows:
BadWindow (invalid Window parameter)

When plotting a figure, editing figure properties and moving to another submenu
(from figure to axes for example). The screen also disappears for a little
while.

(In reply to comment #21)
> Patches work.
> 

Cool and thanks for testing them! 
I'll commit them later and will then close this bug since
the x86_64 issues seem to be resolved.

> Out of curiosity, do you get (if using the tcltk interface):
> 
> Scilab : X error trapped - error message follows:
> BadWindow (invalid Window parameter)
> 
> When plotting a figure, editing figure properties and moving to another submenu
> (from figure to axes for example). The screen also disappears for a little
> while.
> 

I haven't tried the tcl/tk interface in a long time hence I couldn't 
say at the moment. Try re-emerging tcl/tk and if this doesn't
help please file a new bug.

Thanks in advance.

Markus

The patches are in portage cvs now. Thanks for you help.

Best,
Markus

Maybe add a ~amd64 keyword?