sage 2.8.13: g0n - cremona.spkg failed to build, 32bit linux, gcc-4.2.2

[Andrzej Giniewicz]

Hi,

I seen in announcement that it fails for "cremona.spkg fails to
compile on Linux/Itanium with older gcc and also Solaris." - but my
system doesn't fall to that category I think so I decided to report...
not counting plenty (281) warnings in file curvesort.cc (about
deprecated conversion from string constant to 'char*') I also get
WHOLE lot (about 8000) errors all in one nature:

../g0n/curvesort.cc:106: error: jump to case label
../g0n/curvesort.cc:105: error: crosses initialization of 'int
<anonymous>[3]'

with different numbers only... problems starts from:

g++ -c -fPIC -g -O2 -DNEW_OP_ORDER -DUSE_PARI_FACTORING -I../include -
DNTL_ALL -I/opt/sage/local/include -I/opt/sage/local/include tsat3.cc
-o tsat3_n.o
In file included from tsat3.cc:37:
../g0n/curvesort.cc ....... etc etc

I think there is no sense to attach such big report... I'm running
current Arch Linux, that is GCC 4.2.2, gLibc 2.7, kernel 2.6.23.8. Is
there some workaround?

anyway thanks for great software, can't wait till there will be R
interface - it will for sure make Sage the software of choice for
many :)
best regards,
Andrzej Giniewicz.

[mabshoff]

Andrzej Giniewicz wrote:
> Hi,
>

Hello Andrzej

> I seen in announcement that it fails for "cremona.spkg fails to
> compile on Linux/Itanium with older gcc and also Solaris." - but my
> system doesn't fall to that category I think so I decided to report...

Excellent that you report this. It seems that the code needs a little
more exposure to different build envs, but that it what Sage does,
i.e. shake out all the build issue bugs rather quickly.

> not counting plenty (281) warnings in file curvesort.cc (about
> deprecated conversion from string constant to 'char*') I also get
> WHOLE lot (about 8000) errors all in one nature:
>
> ../g0n/curvesort.cc:106: error: jump to case label
> ../g0n/curvesort.cc:105: error: crosses initialization of 'int
> <anonymous>[3]'
>
> with different numbers only... problems starts from:
>
> g++ -c -fPIC -g -O2 -DNEW_OP_ORDER -DUSE_PARI_FACTORING -I../include -
> DNTL_ALL -I/opt/sage/local/include -I/opt/sage/local/include tsat3.cc
> -o tsat3_n.o
> In file included from tsat3.cc:37:
> ../g0n/curvesort.cc ....... etc etc
>
> I think there is no sense to attach such big report... I'm running
> current Arch Linux, that is GCC 4.2.2, gLibc 2.7, kernel 2.6.23.8.

Please move install.log somewhere else and execute make again. Then
gzip up the resulting install.log and put it up somewhere on the web
and send us the link where we can find it.

>Is there some workaround?

Do a "touch spkg/installed/cremona-20071116.p0" from $SAGE_ROOT

Execute make again. Then when building sage.spkg fails go into into
$SAGE_ROOT/devel/sage/setup.py and uncomment lines 524, 525, i.e.

cremona_mat, \
cremona_homspace, \

Then do "sage -b", after that "touch spkg/installed/sage-2.8.13" and
then execute make again.

> anyway thanks for great software, can't wait till there will be R
> interface - it will for sure make Sage the software of choice for
> many :)

Let's hope that happens soon.

> best regards,
> Andrzej Giniewicz.

Cheers,

Michael

[mabshoff]

Sorry to reply to myself this quickly, but this issue is now ticket
#1247, see

http://www.sagetrac.org/sage_trac/ticket/1247

Cheers,

Michael

[Gregory Vanuxem]

Hello,

Le vendredi 23 novembre 2007 à 04:47 -0800, mabshoff a écrit :

[...]

> >
> > g++ -c -fPIC -g -O2 -DNEW_OP_ORDER -DUSE_PARI_FACTORING -I../include -
> > DNTL_ALL -I/opt/sage/local/include -I/opt/sage/local/include tsat3.cc
> > -o tsat3_n.o
> > In file included from tsat3.cc:37:
> > ../g0n/curvesort.cc ....... etc etc
> >
> > I think there is no sense to attach such big report... I'm running
> > current Arch Linux, that is GCC 4.2.2, gLibc 2.7, kernel 2.6.23.8.
>
> Please move install.log somewhere else and execute make again. Then
> gzip up the resulting install.log and put it up somewhere on the web
> and send us the link where we can find it.

I encountered the same issue on Debian testing (amd64/gcc-4.2.3) and had
to switch to gcc-4.1.

Hope that helps,

Greg

[mabshoff]

On Nov 23, 2:27 pm, Gregory Vanuxem <g.vanu...@orange.fr> wrote:
> Hello,
>
> Le vendredi 23 novembre 2007 à 04:47 -0800, mabshoff a écrit :
>
> [...]
>
>
>
> > > g++ -c -fPIC -g -O2 -DNEW_OP_ORDER -DUSE_PARI_FACTORING -I../include -
> > > DNTL_ALL -I/opt/sage/local/include -I/opt/sage/local/include tsat3.cc
> > > -o tsat3_n.o
> > > In file included from tsat3.cc:37:
> > > ../g0n/curvesort.cc ....... etc etc
>
> > > I think there is no sense to attach such big report... I'm running
> > > current Arch Linux, that is GCC 4.2.2, gLibc 2.7, kernel 2.6.23.8.
>
> > Please move install.log somewhere else and execute make again. Then
> > gzip up the resulting install.log and put it up somewhere on the web
> > and send us the link where we can find it.

Hello Greg,

>
> I encountered the same issue on Debian testing (amd64/gcc-4.2.3) and had
> to switch to gcc-4.1.

It seems to boil down to a fcc 4.2.x problem - I will investigate this
some time tonight. John, are you reading sage-support and are aware of
this issue?

> Hope that helps,

Yep, it certainly looks like it.

> Greg

Cheers,

Michael

[William Stein]

On Nov 22, 2007 7:43 AM, Andrzej Giniewicz <ggi...@gmail.com> wrote:
>
> Hi,
>
> I seen in announcement that it fails for "cremona.spkg fails to
> compile on Linux/Itanium with older gcc and also Solaris." - but my
> system doesn't fall to that category I think so I decided to report...
> not counting plenty (281) warnings in file curvesort.cc (about
> deprecated conversion from string constant to 'char*') I also get
> WHOLE lot (about 8000) errors all in one nature:
>
> ../g0n/curvesort.cc:106: error: jump to case label
> ../g0n/curvesort.cc:105: error: crosses initialization of 'int
> <anonymous>[3]'
>
> with different numbers only... problems starts from:
>
> g++ -c -fPIC -g -O2 -DNEW_OP_ORDER -DUSE_PARI_FACTORING -I../include -
> DNTL_ALL -I/opt/sage/local/include -I/opt/sage/local/include tsat3.cc
> -o tsat3_n.o
> In file included from tsat3.cc:37:
> ../g0n/curvesort.cc ....... etc etc
>
> I think there is no sense to attach such big report... I'm running
> current Arch Linux, that is GCC 4.2.2, gLibc 2.7, kernel 2.6.23.8. Is
> there some workaround?

Unfortunately, right now I think the easiest work-around would be
to use GCC 4.1.x instead. If you do try this, please report back
on the results.

> anyway thanks for great software, can't wait till there will be R
> interface - it will for sure make Sage the software of choice for
> many :)

In theory one should just be able to install rpy into sage, so
this shouldn't be too far off...
http://rpy.sourceforge.net/
However, to install rpy, one needs an install of r built in a certain
way to be available on your system. I'm not suggesting you
try this, but just that most of the work is already done.

William

[John Cremona]

I am currently making sure that my code compiles ok with gcc 4.2.1 and
will upload a patch so this will be fixed soon.

John Cremona

--
John Cremona

[mabshoff]

On Nov 23, 9:15 pm, "John Cremona" <john.crem...@gmail.com> wrote:
> I am currently making sure that my code compiles ok with gcc 4.2.1 and
> will upload a patch so this will be fixed soon.
>
> John Cremona

Ok, an updated spkg that merges John's changes is at

http://sage.math.washington.edu/home/mabshoff/cremona-20071124.spkg

I had some slight merge conflicts that I did resolve. It builds an
passes doctests on sage.math. Please try this on systems with gcc
4.2.x and report back.

Cheers,

Michael

[Andrzej Giniewicz]

Great work, that error is no longer here :)... I'm still compiling
other packages so if there will be any further problems I will report
back, but it is already after last time critical point...

sorry for long reply and no log, I was really busy with my tasks and
today checked list first time from posting...

best regards,
Andrzej.

On 24 Lis, 04:02, mabshoff <Michael.Absh...@fsmath.mathematik.uni-

[Simon King]

> Ok, an updated spkg that merges John's changes is at
>
> http://sage.math.washington.edu/home/mabshoff/cremona-20071124.spkg
>
> I had some slight merge conflicts that I did resolve. It builds an
> passes doctests on sage.math. Please try this on systems with gcc
> 4.2.x and report back.

It works for me with gcc 4.2.1.
Thank you!

Yours
Simon