Sage 7.1 Windows 8.1 Cygwin make error

[Witold Żarowski]

I'm trying to install Sage on Windows 8.1/Cygwin.
I'm getting the following error after make command (found in mpfr-3.1.4.log):
===============================
In file included from exceptions.c:23:0:
mpfr-impl.h:261:4: error: #error "__unix__ is defined and __GMP_LIBGMP_DLL is true"
 #  error "__unix__ is defined and __GMP_LIBGMP_DLL is true"
    ^
mpfr-impl.h:264:4: error: #error "Both __unix__ and MPFR_WIN_THREAD_SAFE_DLL are defined"
 #  error "Both __unix__ and MPFR_WIN_THREAD_SAFE_DLL are defined"
    ^
===============================

[Dima Pasechnik]

apparently mpfr needs a cygwin-specific patch:

https://cygwin.com/cgi-bin2/package-cat.cgi?file=x86_64%2Fmpfr%2Fmpfr-3.1.4-1-src&grep=mpfr

and it seems that Sage does not contain it.

(Sage/cygwin used to work, with an older mpfr, but then an update was done, without any cygwin-specific patches...)

[Witold Żarowski]

On Saturday, March 26, 2016 at 5:36:02 PM UTC+1, Dima Pasechnik wrote:

apparently mpfr needs a cygwin-specific patch:

https://cygwin.com/cgi-bin2/package-cat.cgi?file=x86_64%2Fmpfr%2Fmpfr-3.1.4-1-src&grep=mpfr

and it seems that Sage does not contain it.

I see.
But what should I do exactly to make Sage compile?

[Erik Bray]

Technically, you'll need to download the Cygwin package for mpfr,
extract it, then take the patch that Dima linked to into and copy it
into build/pkgs/mpfr/patches (which would have to be created). It was
actually removed relatively recently in
http://git.sagemath.org/sage.git/commit/?id=5e534dcda6ad0b803d5db6b115cc4445721d231a
since the only patch that we did have for mpfr was taken out. So you
would have to modify the spkg-install script in build/pkgs/mpfr to
apply this patch (preferably only when building on Cygwin?).

It looks like a ticket hasn't been made for this yet either, though
there should be. I can try to work in this since I'm working on
building sage in Cygwin anyways.

[Witold Żarowski]

Thanks. I could theoretically do it, if I only knew
1) where to get the Cygwin package for mpfr from and
2) how exactly to change the spkg-install script.

[Dima Pasechnik]

On Tuesday, March 29, 2016 at 6:49:27 PM UTC+1, Witold Żarowski wrote:

Thanks. I could theoretically do it, if I only knew
1) where to get the Cygwin package for mpfr from and

probably you need this 2-line patch:

https://github.com/Alexpux/MSYS2-packages/blob/master/mpfr/mpfr-3.1.4-1.src.patch

(this is perhaps not quite the same as the "proper" cygwin patch - MSYS2 is a fork of Cygwin, 

which is sort of more developer-friendly --- perhaps we should consider switching to it)

 

2) how exactly to change the spkg-install script.

look for examples in directories of other packages.

Usually you put the patch into patches/ directory in build/pkgs/mpfr/

and call patch with appropriate parameters in spkg-install.

[Witold Żarowski]

I can confirm that I was able to compile mprf using the patch you provided.
Then ATLAS was compiling for ages, and after that another error occurred, this time compiling brial:
========================================
libtool: compile:  g++ -DHAVE_CONFIG_H -I. -I../../libpolybori/include/polybori -I../../groebner/include -I../../libpolybori/include -std=c++98 -ftemplate-depth-100 -g -O2 -MT BlockDegLexOrder.lo -MD -MP -MF .deps/BlockDegLexOrder.Tpo -c BlockDegLexOrder.cc  -DDLL_EXPORT -DPIC -o .libs/BlockDegLexOrder.o
In file included from ../../libpolybori/include/polybori/cudd/../../../../Cudd/cudd/cudd.h:90:0,
                 from ../../libpolybori/include/polybori/cudd/cudd.h:20,
                 from ../../libpolybori/include/polybori/pbori_defs.h:27,
                 from ../../libpolybori/include/polybori/BlockDegLexOrder.h:19,
                 from BlockDegLexOrder.cc:18:
/usr/lib/gcc/x86_64-pc-cygwin/5.3.0/include/c++/cstdio:175:11: error: '::snprintf' has not been declared
   using ::snprintf;
           ^
/usr/lib/gcc/x86_64-pc-cygwin/5.3.0/include/c++/cstdio:176:11: error: '::vfscanf' has not been declared
   using ::vfscanf;
           ^
========================================

[Erik Bray]

On Tue, Mar 29, 2016 at 8:41 PM, Dima Pasechnik <dim...@gmail.com> wrote:
>
>
> On Tuesday, March 29, 2016 at 6:49:27 PM UTC+1, Witold Żarowski wrote:
>>
>> Thanks. I could theoretically do it, if I only knew
>> 1) where to get the Cygwin package for mpfr from and
>
>
> probably you need this 2-line patch:
> https://github.com/Alexpux/MSYS2-packages/blob/master/mpfr/mpfr-3.1.4-1.src.patch
> (this is perhaps not quite the same as the "proper" cygwin patch - MSYS2 is
> a fork of Cygwin,
> which is sort of more developer-friendly --- perhaps we should consider
> switching to it)

I'm working on that ;)

[alexander.s...@googlemail.com]

I faced the same error, which appears to be caused by gcc-5.3.0. After downgrading to gcc-4.9.3, brial compiled successfully.
I am now stuck with a segmentation fault related to doc-html.

[Dima Pasechnik]

On Tuesday, April 5, 2016 at 7:56:01 AM UTC+1, alexander.s...@googlemail.com wrote:

I faced the same error, which appears to be caused by gcc-5.3.0. After downgrading to gcc-4.9.3, brial compiled successfully.
I am now stuck with a segmentation fault related to doc-html.

It seems to crash on 

"Setting permissions of DOT_SAGE directory so only you can read and write it."

do you have your home directory in an unusual location/with unusual permissions?

(cygwin is not known to be 100% compatible in this sense, e.g. on Samba shares things

can go funny, etc...)

[Francesco Biscani]

On 29 March 2016 at 20:41, Dima Pasechnik <dim...@gmail.com> wrote:

(this is perhaps not quite the same as the "proper" cygwin patch - MSYS2 is a fork of Cygwin, 

which is sort of more developer-friendly --- perhaps we should consider switching to it)

As far as I know MSYS2  is not a fork of cygwin, it is a native mingw toolchain + plus a bunch of natively-compiled packages managed by the package manager of Arch linux, called pacman.

[Dima Pasechnik]

MSYS2 has a  mingw toolchain and a Cygwin toolchain. 

http://stackoverflow.com/questions/25019057/how-are-msys-msys2-and-msysgit-related-to-each-other

[alexander.s...@googlemail.com]

My home directory is in C:\cygwin64. Permission were mixed up, but after restoring them the error persists.
Although sage is no longer complaining about DOT_SAGE permissions.

[Dima Pasechnik]

Does Sage start for you?

[alexander.s...@googlemail.com]

No, start-up fails with a related(?) seg fault.

[Francesco Biscani]

On 5 April 2016 at 10:59, Dima Pasechnik <dim...@gmail.com> wrote:

MSYS2 has a  mingw toolchain and a Cygwin toolchain. 

http://stackoverflow.com/questions/25019057/how-are-msys-msys2-and-msysgit-related-to-each-other

I did not know that msys2 also had a cygwin toolchain. But then, I don't want to touch cygwin with a barge pole so that's hardly surprising :) 

[alexander.s...@googlemail.com]

The failure looks similar to http://trac.sagemath.org/ticket/11551. Any help would be appreciated :)

[Erik Bray]

On Tue, Apr 5, 2016 at 8:53 AM, alexander.stottmeister via sage-devel
<sage-...@googlegroups.com> wrote:
> I faced the same error, which appears to be caused by gcc-5.3.0. After
> downgrading to gcc-4.9.3, brial compiled successfully.
> I am now stuck with a segmentation fault related to doc-html.

I'm starting to work on Cygwin builds now myself. I've gone ahead and
created a ticket + patch for the MPFR issue:

http://trac.sagemath.org/ticket/20380

Now I've hit the same issue with brial. Downgrading gcc is not a
solution as far as I'm concerned so I'll see if I can understand this
issue and find out what's going on. No idea yet about doc-html.

> --
> You received this message because you are subscribed to the Google Groups
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.