[sage-devel] Re: Building binaries...

2016-05-09 Thread Dima Pasechnik
A different (Linux-only ?) approach to this problem might be to use unshare 
(1) and private mounts.
See https://github.com/mwilliamson/whack

 

On Saturday, May 7, 2016 at 4:46:20 PM UTC+1, William wrote:
>
> On Sat, May 7, 2016 at 2:00 AM, Volker Braun  > wrote: 
> > The sage script in your PATH is outdated: 
> > 
> > $ ./sage -advanced | grep bdist 
> > $ egrep -r bdist src/ 
> > $ 
> > 
> > The way to build binary (relocatable) packages is 
> > https://github.com/sagemath/binary-pkg, which says: 
> > 
> > git clone https://github.com/sagemath/binary-pkg.git 
> > cd binary-pkg 
> > make bdist-sage-linux # If you are on Linux 
> > make bdist-sage-osx   # If you are on OSX 
> > ls dist/  # Built binaries will be in this directory 
>
> I read that github page, but I don't know what binary-pkg actually 
> does.  The docs don't answer my questions. 
>
> For context, I used to (1) build a copy of Sage, (2) possibly 
> customize it, then (3) type 
>
>   ./sage -bdist  
>
> to make a binary from it. 
>
> What does this binary-pkg program do?   Does it package up an existing 
> Sage install?  If so, how do you specify which sage install it 
> packages?   Yes, I read over the yaml file but couldn't figure it out. 
>
>
>  
> I am sad that I can't copy existing sage installs, and I'm now even 
> sadder that I can't type "./sage -bdist" anymore.  Two of the most 
> important things for making sage dev easy for people are gone.   If I 
> had the time, I would make a fork of Sage that restored exactly this 
> behavior then always merge it into sage... 
>  
>
> William 
>
>
> > 
> > 
> > 
> > On Saturday, May 7, 2016 at 2:05:00 AM UTC+2, William Stein wrote: 
> >> 
> >> Either my build is completely hosed, or "sage -bdist" is completely 
> >> broken: 
> >> 
> >> salvus@compute7-us:/projects/sage/sage-develop$ time ./sage -bdist tmp 
> >> sage-run received unknown option: -bdist 
> >> usage: sage [options] 
> >> Try 'sage -h' for more information. 
> >> salvus@compute7-us:/projects/sage/sage-develop$ sage -advanced|grep 
> dist 
> >> Making Sage packages or distributions: 
> >>   -bdist  -- build a binary distribution of Sage 
> >>   -sdist  -- build a source distribution of Sage 
> >> salvus@compute7-us:/projects/sage/sage-develop$ ls 
> >> aclocal.m4  bootstrap  config  config.status  configure.ac 
> >> local  m4README.md  src   VERSION.txt 
> >> autom4te.cache  build  config.log  configure  COPYING.txt 
> >> logs   Makefile  sage   upstream 
> >> 
> >> --- 
> >> 
> >> Also, even if bdist worked, I don't understand the docs.  What's 
> >> ??!?? Where does the bdist actually go??   It used to go in a 
> >> directory SAGE_ROOT/dist, I think. 
> >> 
> >> William 
> >> 
> >> 
> >> On Fri, May 6, 2016 at 2:42 AM, Eric Gourgoulhon  
> >> wrote: 
> >> > As it stands, symbolic calculus is quite severely broken in Sage 7.2: 
> >> > making 
> >> > any assumption on a symbolic variable turns it into an integer. For 
> >> > instance 
> >> > assume(x>0) yields sin(pi*x)=0. This has many undesirable 
> consequences 
> >> > and 
> >> > users will probably complain a lot. Fortunately a fix is provided by 
> the 
> >> > upgrade to the latest version of pynac, which is proposed in #20475. 
> I 
> >> > am 
> >> > reviewing this ticket and IMHO it should be merged before releasing 
> 7.2. 
> >> > 
> >> > -- 
> >> > You received this message because you are subscribed to the Google 
> >> > Groups 
> >> > "sage-release" group. 
> >> > To unsubscribe from this group and stop receiving emails from it, 
> send 
> >> > an 
> >> > email to sage-release...@googlegroups.com. 
> >> > To post to this group, send email to sage-r...@googlegroups.com. 
> >> > Visit this group at https://groups.google.com/group/sage-release. 
> >> > For more options, visit https://groups.google.com/d/optout. 
> >> 
> >> 
> >> 
> >> -- 
> >> William (http://wstein.org) 
> > 
> > -- 
> > You received this message because you are subscribed to the Google 
> Groups 
> > "sage-release" group. 
> > To unsubscribe from this group and stop receiving emails from it, send 
> an 
> > email to sage-release...@googlegroups.com . 
> > To post to this group, send email to sage-r...@googlegroups.com 
> . 
> > Visit this group at https://groups.google.com/group/sage-release. 
> > For more options, visit https://groups.google.com/d/optout. 
>
>
>
> -- 
> William (http://wstein.org) 
>

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Jeroen Demeyer
This has been the default for a long time now, there is no need to add 
it explicitly:


On 2016-05-09 00:17, Volker Braun wrote:

 export SAGE_PARALLEL_SPKG_BUILD=yes


--
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Thierry
On Sun, May 08, 2016 at 03:17:44PM -0700, Volker Braun wrote:
> The intended way to customize this is to create your own sage-debian.yaml 
> where you modify the build instructions as you want them:
> 
> build: |
> export SAGE_FAT_BINARY=yes
> export SAGE_PARALLEL_SPKG_BUILD=yes
> export MAKE='make -j{ncpu}'
> make
> git gc --aggressive --prune=now
> 
> 
> Then "make package-sage-debian" will build that.

OK thanks for the hint, i will try that way for the next release.

Ciao
Thierry


> 
> 
> 
> On Sunday, May 8, 2016 at 11:52:20 PM UTC+2, Thierry (sage-googlesucks@xxx) 
> wrote:
> >
> > Hi, 
> >
> > i have to witness that for Sage Debian Live, i currently have to fork the 
> > behaviour of "sage -bdist" to continue maintain the live. Indeed, it is 
> > still not clear to me how to use the new binary-pkg to package a 
> > customized sage install. To let the key be self-contained, i install most 
> > optional packages (+some experimental), and also some popular pip packages 
> > (at least the standard scipy stack) before building the binary tarball. 
> >
> > What are the steps to achieve that with binary-pkg to get a relocatable 
> > customized build ? 
> >
> > Ciao, 
> > Thierry 
> >
> >
> 
> -- 
> 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+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Thierry
On Sun, May 08, 2016 at 03:02:19PM -0700, William Stein wrote:
> On Sunday, May 8, 2016, Thierry  wrote:
> 
> > On Sun, May 08, 2016 at 02:53:58PM -0700, William Stein wrote:
> > > On Sunday, May 8, 2016, Thierry  > > wrote:
> > >
> > > > Hi,
> > > >
> > > > i have to witness that for Sage Debian Live, i currently have to fork
> > the
> > > > behaviour of "sage -bdist" to continue maintain the live. Indeed, it is
> > >
> > >
> > > Where is your fork?
> >
> > Not exactly a fork, more an imitation of the behaviour: i copied the
> > interesting lines of the old sage-bdist file directly to my builder
> > scripts.
> 
> 
> Where are your scripts?

They used to be available at http://sagebuilder.metelu.net/ but i got a
disk crash last month and i only relocated my emails and important
webpages such as http://sagedebianlive.metelu.net/ to some friend
meanwhile. Most should be back online within 2 weeks hopefully.

Basically, those scripts build/updgrade/boot qemu/kvm images and build
Sage binaries within them, one after another. They were used to build some
missing Ubuntu and Debian binaries some time ago, and i still use them for
the live to get a 32 bit system that is bootable on most computers (and my
laptop is 64).

Ciao,
Thierry


> 
> 
> 
> 
> 
> >
> > Ciao,
> > Thierry
> >
> >
> > > > still not clear to me how to use the new binary-pkg to package a
> > > > customized sage install. To let the key be self-contained, i install
> > most
> > > > optional packages (+some experimental), and also some popular pip
> > packages
> > > > (at least the standard scipy stack) before building the binary tarball.
> > > >
> > > > What are the steps to achieve that with binary-pkg to get a relocatable
> > > > customized build ?
> > > >
> > > > Ciao,
> > > > Thierry
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> > > > > On Sunday, May 8, 2016, Volker Braun  > 
> > > > > wrote:
> > > > >
> > > > > > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> > > > > >>
> > > > > >> which is 593M.
> > > > > >>
> > > > > >
> > > > > > Sounds a bit too small, the buildbot binaries are always about
> > twice
> > > > that.
> > > > > >
> > > > >
> > > > >
> > > > > Ok that's good to know and explains why the binaries didn't work.
> > I'll
> > > > try
> > > > > building again and report any log errors.
> > > > >
> > > > >
> > > > > >
> > > > > > instantly prints out massive screenfulls of information -- with no
> > > > > >> warning, and no indicator that hitting control+c is a bad idea.
> > > > > >
> > > > > >
> > > > > > Wait, so you did Ctrl-C the installatiion?
> > > > > >
> > > > > >
> > > > > No but as you noted above my binary is clearly broken.  I'll figure
> > that
> > > > > out first and report back.  Thanks for being so patient and helpful.
> > > > >
> > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > --
> > > > > > 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+unsubscr...@googlegroups.com 
> > 
> > > > > >  > > > sage-devel%2bunsubscr...@googlegroups.com 
> > ');>
> > > > > > .
> > > > > > To post to this group, send email to sage-devel@googlegroups.com
> > 
> > > > 
> > > > > >  > 
> > > > ');>.
> > > > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > > > For more options, visit https://groups.google.com/d/optout.
> > > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Sent from my massive iPhone 6 plus.
> > > > >
> > > > > --
> > > > > 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+unsubscr...@googlegroups.com 
> > .
> > > > > To post to this group, send email to sage-devel@googlegroups.com
> > 
> > > > .
> > > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > > For more options, visit https://groups.google.com/d/optout.
> > > >
> > > > --
> > > > 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+unsubscr...@googlegroups.com 
> > .
> > > > To post to this group, send email to sage-devel@googlegroups.com
> > 
> > > > .
> > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > For more options, visit https://groups.google.com/d/optout.
> > > >
> > >
> > >
> > > --
> > > Sent from my massive iPhone 6 plus.
> > >
> > > --
> > > 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+unsubscr...@googlegroups.com 

Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Volker Braun
The intended way to customize this is to create your own sage-debian.yaml 
where you modify the build instructions as you want them:

build: |
export SAGE_FAT_BINARY=yes
export SAGE_PARALLEL_SPKG_BUILD=yes
export MAKE='make -j{ncpu}'
make
git gc --aggressive --prune=now


Then "make package-sage-debian" will build that.



On Sunday, May 8, 2016 at 11:52:20 PM UTC+2, Thierry (sage-googlesucks@xxx) 
wrote:
>
> Hi, 
>
> i have to witness that for Sage Debian Live, i currently have to fork the 
> behaviour of "sage -bdist" to continue maintain the live. Indeed, it is 
> still not clear to me how to use the new binary-pkg to package a 
> customized sage install. To let the key be self-contained, i install most 
> optional packages (+some experimental), and also some popular pip packages 
> (at least the standard scipy stack) before building the binary tarball. 
>
> What are the steps to achieve that with binary-pkg to get a relocatable 
> customized build ? 
>
> Ciao, 
> Thierry 
>
>

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Thierry
On Sun, May 08, 2016 at 10:08:14PM +0200, Thierry wrote:
> On Sun, May 08, 2016 at 02:53:58PM -0700, William Stein wrote:
> > On Sunday, May 8, 2016, Thierry  wrote:
> > 
> > > Hi,
> > >
> > > i have to witness that for Sage Debian Live, i currently have to fork the
> > > behaviour of "sage -bdist" to continue maintain the live. Indeed, it is
> > 
> > 
> > Where is your fork?
> 
> Not exactly a fork, more an imitation of the behaviour: i copied the
> interesting lines of the old sage-bdist file directly to my builder
> scripts.

Also, because the older relocation stuff is gone, i had to modify the
scripts to build at the exact same place than the target directory.

Ciao,
Thierry


> 
> Ciao,
> Thierry
> 
>  
> > > still not clear to me how to use the new binary-pkg to package a
> > > customized sage install. To let the key be self-contained, i install most
> > > optional packages (+some experimental), and also some popular pip packages
> > > (at least the standard scipy stack) before building the binary tarball.
> > >
> > > What are the steps to achieve that with binary-pkg to get a relocatable
> > > customized build ?
> > >
> > > Ciao,
> > > Thierry
> > >
> > >
> > >
> > >
> > >
> > > On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> > > > On Sunday, May 8, 2016, Volker Braun  > > > wrote:
> > > >
> > > > > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> > > > >>
> > > > >> which is 593M.
> > > > >>
> > > > >
> > > > > Sounds a bit too small, the buildbot binaries are always about twice
> > > that.
> > > > >
> > > >
> > > >
> > > > Ok that's good to know and explains why the binaries didn't work.  I'll
> > > try
> > > > building again and report any log errors.
> > > >
> > > >
> > > > >
> > > > > instantly prints out massive screenfulls of information -- with no
> > > > >> warning, and no indicator that hitting control+c is a bad idea.
> > > > >
> > > > >
> > > > > Wait, so you did Ctrl-C the installatiion?
> > > > >
> > > > >
> > > > No but as you noted above my binary is clearly broken.  I'll figure that
> > > > out first and report back.  Thanks for being so patient and helpful.
> > > >
> > > >
> > > > >
> > > > >
> > > > >
> > > > --
> > > > > 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+unsubscr...@googlegroups.com 
> > > > >  > > sage-devel%2bunsubscr...@googlegroups.com ');>
> > > > > .
> > > > > To post to this group, send email to sage-devel@googlegroups.com
> > > 
> > > > >  > > ');>.
> > > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > > For more options, visit https://groups.google.com/d/optout.
> > > > >
> > > >
> > > >
> > > > --
> > > > Sent from my massive iPhone 6 plus.
> > > >
> > > > --
> > > > 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+unsubscr...@googlegroups.com .
> > > > To post to this group, send email to sage-devel@googlegroups.com
> > > .
> > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > For more options, visit https://groups.google.com/d/optout.
> > >
> > > --
> > > 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+unsubscr...@googlegroups.com .
> > > To post to this group, send email to sage-devel@googlegroups.com
> > > .
> > > Visit this group at https://groups.google.com/group/sage-devel.
> > > For more options, visit https://groups.google.com/d/optout.
> > >
> > 
> > 
> > -- 
> > Sent from my massive iPhone 6 plus.
> > 
> > -- 
> > 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+unsubscr...@googlegroups.com.
> > To post to this group, send email to sage-devel@googlegroups.com.
> > Visit this group at https://groups.google.com/group/sage-devel.
> > For more options, visit https://groups.google.com/d/optout.
> 
> -- 
> 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+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

-- 
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+unsubscr...@googleg

[sage-devel] Re: Building binaries...

2016-05-08 Thread William Stein
On Sunday, May 8, 2016, Thierry  wrote:

> On Sun, May 08, 2016 at 02:53:58PM -0700, William Stein wrote:
> > On Sunday, May 8, 2016, Thierry  > wrote:
> >
> > > Hi,
> > >
> > > i have to witness that for Sage Debian Live, i currently have to fork
> the
> > > behaviour of "sage -bdist" to continue maintain the live. Indeed, it is
> >
> >
> > Where is your fork?
>
> Not exactly a fork, more an imitation of the behaviour: i copied the
> interesting lines of the old sage-bdist file directly to my builder
> scripts.


Where are your scripts?





>
> Ciao,
> Thierry
>
>
> > > still not clear to me how to use the new binary-pkg to package a
> > > customized sage install. To let the key be self-contained, i install
> most
> > > optional packages (+some experimental), and also some popular pip
> packages
> > > (at least the standard scipy stack) before building the binary tarball.
> > >
> > > What are the steps to achieve that with binary-pkg to get a relocatable
> > > customized build ?
> > >
> > > Ciao,
> > > Thierry
> > >
> > >
> > >
> > >
> > >
> > > On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> > > > On Sunday, May 8, 2016, Volker Braun  
> > > > wrote:
> > > >
> > > > > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> > > > >>
> > > > >> which is 593M.
> > > > >>
> > > > >
> > > > > Sounds a bit too small, the buildbot binaries are always about
> twice
> > > that.
> > > > >
> > > >
> > > >
> > > > Ok that's good to know and explains why the binaries didn't work.
> I'll
> > > try
> > > > building again and report any log errors.
> > > >
> > > >
> > > > >
> > > > > instantly prints out massive screenfulls of information -- with no
> > > > >> warning, and no indicator that hitting control+c is a bad idea.
> > > > >
> > > > >
> > > > > Wait, so you did Ctrl-C the installatiion?
> > > > >
> > > > >
> > > > No but as you noted above my binary is clearly broken.  I'll figure
> that
> > > > out first and report back.  Thanks for being so patient and helpful.
> > > >
> > > >
> > > > >
> > > > >
> > > > >
> > > > --
> > > > > 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+unsubscr...@googlegroups.com 
> 
> > > > >  > > sage-devel%2bunsubscr...@googlegroups.com 
> ');>
> > > > > .
> > > > > To post to this group, send email to sage-devel@googlegroups.com
> 
> > > 
> > > > >  
> > > ');>.
> > > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > > For more options, visit https://groups.google.com/d/optout.
> > > > >
> > > >
> > > >
> > > > --
> > > > Sent from my massive iPhone 6 plus.
> > > >
> > > > --
> > > > 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+unsubscr...@googlegroups.com 
> .
> > > > To post to this group, send email to sage-devel@googlegroups.com
> 
> > > .
> > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > For more options, visit https://groups.google.com/d/optout.
> > >
> > > --
> > > 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+unsubscr...@googlegroups.com 
> .
> > > To post to this group, send email to sage-devel@googlegroups.com
> 
> > > .
> > > Visit this group at https://groups.google.com/group/sage-devel.
> > > For more options, visit https://groups.google.com/d/optout.
> > >
> >
> >
> > --
> > Sent from my massive iPhone 6 plus.
> >
> > --
> > 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+unsubscr...@googlegroups.com .
> > To post to this group, send email to sage-devel@googlegroups.com
> .
> > Visit this group at https://groups.google.com/group/sage-devel.
> > For more options, visit https://groups.google.com/d/optout.
>
> --
> 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+unsubscr...@googlegroups.com .
> To post to this group, send email to sage-devel@googlegroups.com
> .
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.
>


-- 
Sent from my massive iPhone 6 plus.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googl

Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Thierry
On Sun, May 08, 2016 at 02:53:58PM -0700, William Stein wrote:
> On Sunday, May 8, 2016, Thierry  wrote:
> 
> > Hi,
> >
> > i have to witness that for Sage Debian Live, i currently have to fork the
> > behaviour of "sage -bdist" to continue maintain the live. Indeed, it is
> 
> 
> Where is your fork?

Not exactly a fork, more an imitation of the behaviour: i copied the
interesting lines of the old sage-bdist file directly to my builder
scripts.

Ciao,
Thierry

 
> > still not clear to me how to use the new binary-pkg to package a
> > customized sage install. To let the key be self-contained, i install most
> > optional packages (+some experimental), and also some popular pip packages
> > (at least the standard scipy stack) before building the binary tarball.
> >
> > What are the steps to achieve that with binary-pkg to get a relocatable
> > customized build ?
> >
> > Ciao,
> > Thierry
> >
> >
> >
> >
> >
> > On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> > > On Sunday, May 8, 2016, Volker Braun  > > wrote:
> > >
> > > > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> > > >>
> > > >> which is 593M.
> > > >>
> > > >
> > > > Sounds a bit too small, the buildbot binaries are always about twice
> > that.
> > > >
> > >
> > >
> > > Ok that's good to know and explains why the binaries didn't work.  I'll
> > try
> > > building again and report any log errors.
> > >
> > >
> > > >
> > > > instantly prints out massive screenfulls of information -- with no
> > > >> warning, and no indicator that hitting control+c is a bad idea.
> > > >
> > > >
> > > > Wait, so you did Ctrl-C the installatiion?
> > > >
> > > >
> > > No but as you noted above my binary is clearly broken.  I'll figure that
> > > out first and report back.  Thanks for being so patient and helpful.
> > >
> > >
> > > >
> > > >
> > > >
> > > --
> > > > 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+unsubscr...@googlegroups.com 
> > > >  > sage-devel%2bunsubscr...@googlegroups.com ');>
> > > > .
> > > > To post to this group, send email to sage-devel@googlegroups.com
> > 
> > > >  > ');>.
> > > > Visit this group at https://groups.google.com/group/sage-devel.
> > > > For more options, visit https://groups.google.com/d/optout.
> > > >
> > >
> > >
> > > --
> > > Sent from my massive iPhone 6 plus.
> > >
> > > --
> > > 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+unsubscr...@googlegroups.com .
> > > To post to this group, send email to sage-devel@googlegroups.com
> > .
> > > Visit this group at https://groups.google.com/group/sage-devel.
> > > For more options, visit https://groups.google.com/d/optout.
> >
> > --
> > 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+unsubscr...@googlegroups.com .
> > To post to this group, send email to sage-devel@googlegroups.com
> > .
> > Visit this group at https://groups.google.com/group/sage-devel.
> > For more options, visit https://groups.google.com/d/optout.
> >
> 
> 
> -- 
> Sent from my massive iPhone 6 plus.
> 
> -- 
> 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+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


[sage-devel] Re: Building binaries...

2016-05-08 Thread William Stein
On Sunday, May 8, 2016, Thierry  wrote:

> Hi,
>
> i have to witness that for Sage Debian Live, i currently have to fork the
> behaviour of "sage -bdist" to continue maintain the live. Indeed, it is


Where is your fork?




> still not clear to me how to use the new binary-pkg to package a
> customized sage install. To let the key be self-contained, i install most
> optional packages (+some experimental), and also some popular pip packages
> (at least the standard scipy stack) before building the binary tarball.
>
> What are the steps to achieve that with binary-pkg to get a relocatable
> customized build ?
>
> Ciao,
> Thierry
>
>
>
>
>
> On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> > On Sunday, May 8, 2016, Volker Braun  > wrote:
> >
> > > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> > >>
> > >> which is 593M.
> > >>
> > >
> > > Sounds a bit too small, the buildbot binaries are always about twice
> that.
> > >
> >
> >
> > Ok that's good to know and explains why the binaries didn't work.  I'll
> try
> > building again and report any log errors.
> >
> >
> > >
> > > instantly prints out massive screenfulls of information -- with no
> > >> warning, and no indicator that hitting control+c is a bad idea.
> > >
> > >
> > > Wait, so you did Ctrl-C the installatiion?
> > >
> > >
> > No but as you noted above my binary is clearly broken.  I'll figure that
> > out first and report back.  Thanks for being so patient and helpful.
> >
> >
> > >
> > >
> > >
> > --
> > > 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+unsubscr...@googlegroups.com 
> > >  sage-devel%2bunsubscr...@googlegroups.com ');>
> > > .
> > > To post to this group, send email to sage-devel@googlegroups.com
> 
> > >  ');>.
> > > Visit this group at https://groups.google.com/group/sage-devel.
> > > For more options, visit https://groups.google.com/d/optout.
> > >
> >
> >
> > --
> > Sent from my massive iPhone 6 plus.
> >
> > --
> > 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+unsubscr...@googlegroups.com .
> > To post to this group, send email to sage-devel@googlegroups.com
> .
> > Visit this group at https://groups.google.com/group/sage-devel.
> > For more options, visit https://groups.google.com/d/optout.
>
> --
> 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+unsubscr...@googlegroups.com .
> To post to this group, send email to sage-devel@googlegroups.com
> .
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.
>


-- 
Sent from my massive iPhone 6 plus.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Thierry
Hi,

i have to witness that for Sage Debian Live, i currently have to fork the
behaviour of "sage -bdist" to continue maintain the live. Indeed, it is
still not clear to me how to use the new binary-pkg to package a
customized sage install. To let the key be self-contained, i install most
optional packages (+some experimental), and also some popular pip packages
(at least the standard scipy stack) before building the binary tarball.

What are the steps to achieve that with binary-pkg to get a relocatable
customized build ?

Ciao,
Thierry





On Sun, May 08, 2016 at 12:00:41PM -0700, William Stein wrote:
> On Sunday, May 8, 2016, Volker Braun  wrote:
> 
> > On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
> >>
> >> which is 593M.
> >>
> >
> > Sounds a bit too small, the buildbot binaries are always about twice that.
> >
> 
> 
> Ok that's good to know and explains why the binaries didn't work.  I'll try
> building again and report any log errors.
> 
> 
> >
> > instantly prints out massive screenfulls of information -- with no
> >> warning, and no indicator that hitting control+c is a bad idea.
> >
> >
> > Wait, so you did Ctrl-C the installatiion?
> >
> >
> No but as you noted above my binary is clearly broken.  I'll figure that
> out first and report back.  Thanks for being so patient and helpful.
> 
> 
> >
> >
> >
> -- 
> > 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+unsubscr...@googlegroups.com
> > 
> > .
> > To post to this group, send email to sage-devel@googlegroups.com
> > .
> > Visit this group at https://groups.google.com/group/sage-devel.
> > For more options, visit https://groups.google.com/d/optout.
> >
> 
> 
> -- 
> Sent from my massive iPhone 6 plus.
> 
> -- 
> 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+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


[sage-devel] Re: Building binaries...

2016-05-08 Thread William Stein
On Sunday, May 8, 2016, Volker Braun  wrote:

> On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
>>
>> which is 593M.
>>
>
> Sounds a bit too small, the buildbot binaries are always about twice that.
>


Ok that's good to know and explains why the binaries didn't work.  I'll try
building again and report any log errors.


>
> instantly prints out massive screenfulls of information -- with no
>> warning, and no indicator that hitting control+c is a bad idea.
>
>
> Wait, so you did Ctrl-C the installatiion?
>
>
No but as you noted above my binary is clearly broken.  I'll figure that
out first and report back.  Thanks for being so patient and helpful.


>
>
>
-- 
> 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+unsubscr...@googlegroups.com
> 
> .
> To post to this group, send email to sage-devel@googlegroups.com
> .
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.
>


-- 
Sent from my massive iPhone 6 plus.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Volker Braun
On Sunday, May 8, 2016 at 6:19:17 PM UTC+2, William wrote:
>
> > b) modifying a shared library in the destination /tmp/sage-dev/ has no 
> > effect until you relink the binary/cython extension that uses it. 
>
> Can you elaborate on b slightly?  I don't understand for sure. 
>


Say, you go to the copy /tmp/sage-dev and then

1) check out a branch with a different mpir verision
2) just install mpir with sage -p mpir

then sage will still use the libmpir.so from /projects. To use the changed 
library you'll have to recompile all dependencies (by running make, or sage 
-f mpir). Until then the rpaths point back to /projects

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread William Stein
On Sun, May 8, 2016 at 1:34 AM, Volker Braun  wrote:
> On Sunday, May 8, 2016 at 6:54:38 AM UTC+2, William wrote:
>>
>> time rsync -axH /projects/sage/sage-dev/  sage-dev/
>
>
> Thats fine. The only caveat is that rpaths are still pointing to the source.
> Therefore
> a) the copy stops working if you delete /projects/sage/sage-dev/

OK, so it's critical that I have a separate build for each version
people might copy, e.g., instead do

  /projects/sage/sage-dev/sage-7.2.rc1

That will also be good for other reasons.

> b) modifying a shared library in the destination /tmp/sage-dev/ has no
> effect until you relink the binary/cython extension that uses it.

Can you elaborate on b slightly?  I don't understand for sure.

 -- William

>
> --
> 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+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.



-- 
William (http://wstein.org)

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Volker Braun
On Sunday, May 8, 2016 at 6:54:38 AM UTC+2, William wrote:
>
> time rsync -axH /projects/sage/sage-dev/  sage-dev/ 
>

Thats fine. The only caveat is that rpaths are still pointing to the 
source. Therefore
a) the copy stops working if you delete /projects/sage/sage-dev/
b) modifying a shared library in the destination /tmp/sage-dev/ has no 
effect until you relink the binary/cython extension that uses it.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-08 Thread Volker Braun
On Sunday, May 8, 2016 at 6:19:43 AM UTC+2, William wrote:
>
> which is 593M. 
>

Sounds a bit too small, the buildbot binaries are always about twice that.

instantly prints out massive screenfulls of information -- with no 
> warning, and no indicator that hitting control+c is a bad idea.


Wait, so you did Ctrl-C the installation?

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-07 Thread William Stein
On Sat, May 7, 2016 at 12:18 PM, Volker Braun  wrote:
> You can set it yourself in the environment after moving Sage, this may work
> with the aforementioned caveats about conflicts with system libraries. And
> change SAGE_LOCAL/lib/sage-current-location.txt to avoid the relocation
> error message. But its definitely not the best solution.
>
> In any case, I don't think saving an hour or two to serve this very special
> usage is worth special casing how we build binaries.

Given an existing Sage install in  /projects/sage/sage-dev/  built in
the standard way, we have, doing the following:

time rsync -axH /projects/sage/sage-dev/  sage-dev/
cd sage-dev
echo `pwd` > local/lib/sage-current-location.txt
./sage -br  # starts quickly (but singular warning)

takes about 3 minutes (on SMC).It results in a copy of Sage that
seems to work fine for development (not having to even wait for the
first 'sage -br'), though there is one single warning displayed when
starting up about Singular:

salvus@compute7-us:/tmp/sage-dev$ ./sage
┌┐
│ SageMath version 7.2.rc1, Release Date: 2016-05-07 │
│ Type "notebook()" for the browser-based notebook interface.│
│ Type "help()" for help.│
└┘
┏┓
┃ Warning: this is a prerelease version, and it may be unstable. ┃
┗┛
// ** Could not open dynamic library: /tmp/sage-dev/local/bin/p_Procs_FieldQ.so
// ** Error message from system:
/tmp/sage-dev/local/bin/p_Procs_FieldQ.so: undefined symbol: nDivBy
// ** Singular will work properly, but much slower.
// ** See the INSTALL section in the Singular manual for details.


I tested things like modifying Cython code, typing "sage -br", etc.,
and dragons did not rain fire down on me.

 -- William


P.S.

Doctesting is "disturbing" since Sage now always allocates 25GB of
virtual memory, which is bizarre:

25626 salvus20   0 24.504g 243544  52592 R  99.3  0.8   0:19.00 python

This happens with every Sage process.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-07 Thread William Stein
Hi,

I just tested this package for creating binaries.   It creates by default a file

  sage-7.2.rc1-Ubuntu_15.10-x86_64.tar.bz2

which is 593M.

Doing

  time tar jxf sage-7.2.rc1-Ubuntu_15.10-x86_64.tar.bz2

takes just under 2 minutes (for me) and results in a directory
SageMath, rather than sage-7.2.rc1-Ubuntu_15.10-x86_64, which is I
think the typical convension for tarballs (namely, extract foo.tar.bz2
to foo). This is annoying, and another change from how binaries used
to work.   Score -1 for this.

The directory SageMath has size 2.1GB, which is really nice actually!
I expected something much bigger.   It also has 49702 files.  Score +1
for this.

Changing to the SageMath directory and doing

   time ./sage

instantly prints out massive screenfulls of information -- with no
warning, and no indicator that hitting control+c is a bad idea.  Score
-1 for this.

After 10 seconds it finishes with

...
patching /tmp/SageMath/local/lib/R/modules/lapack.so
patching 
/tmp/SageMath/local/gap/gap-4.8.3/bin/x86_64-unknown-linux-gnu-gcc-default64/iostream.o

It seems that you are attempting to run Sage from an unpacked source
tarball, but you have not compiled it yet (or maybe the build has not
finished). You should run `make` in the Sage root directory first.
If you did not intend to build Sage from source, you should download
a binary tarball instead. Read README.txt for more information.


real0m10.885s
user0m0.740s
sys 0m0.360s
/tmp/SageMath$

So, as far as I can tell, complete fail.  Score -oo for this.

This is all on SageMathCloud in /tmp, and anybody should be able to
replicate this.

 -- William



On Sat, May 7, 2016 at 9:02 AM, Dima Pasechnik  wrote:
> Yes, I know, I was puzzled by this some time ago too. The reason for all
> this mess is that in order to make
> a relocatable Sage binary (more precisely, a bunch of relocateable libs etc)
> that uses rpath, it has to be built at a location that will allow
> pattern-matching on binaries to work (so it has to be some long and
> convoluted string).
> Thus your usual directory where you build Sage won't work.
>
> Dima
>
> On Saturday, May 7, 2016 at 4:46:20 PM UTC+1, William wrote:
>>
>> On Sat, May 7, 2016 at 2:00 AM, Volker Braun  wrote:
>> > The sage script in your PATH is outdated:
>> >
>> > $ ./sage -advanced | grep bdist
>> > $ egrep -r bdist src/
>> > $
>> >
>> > The way to build binary (relocatable) packages is
>> > https://github.com/sagemath/binary-pkg, which says:
>> >
>> > git clone https://github.com/sagemath/binary-pkg.git
>> > cd binary-pkg
>> > make bdist-sage-linux # If you are on Linux
>> > make bdist-sage-osx   # If you are on OSX
>> > ls dist/  # Built binaries will be in this directory
>>
>> I read that github page, but I don't know what binary-pkg actually
>> does.  The docs don't answer my questions.
>>
>> For context, I used to (1) build a copy of Sage, (2) possibly
>> customize it, then (3) type
>>
>>   ./sage -bdist 
>>
>> to make a binary from it.
>>
>> What does this binary-pkg program do?   Does it package up an existing
>> Sage install?  If so, how do you specify which sage install it
>> packages?   Yes, I read over the yaml file but couldn't figure it out.
>>
>>
>> 
>> I am sad that I can't copy existing sage installs, and I'm now even
>> sadder that I can't type "./sage -bdist" anymore.  Two of the most
>> important things for making sage dev easy for people are gone.   If I
>> had the time, I would make a fork of Sage that restored exactly this
>> behavior then always merge it into sage...
>> 
>>
>> William
>>
>>
>> >
>> >
>> >
>> > On Saturday, May 7, 2016 at 2:05:00 AM UTC+2, William Stein wrote:
>> >>
>> >> Either my build is completely hosed, or "sage -bdist" is completely
>> >> broken:
>> >>
>> >> salvus@compute7-us:/projects/sage/sage-develop$ time ./sage -bdist tmp
>> >> sage-run received unknown option: -bdist
>> >> usage: sage [options]
>> >> Try 'sage -h' for more information.
>> >> salvus@compute7-us:/projects/sage/sage-develop$ sage -advanced|grep
>> >> dist
>> >> Making Sage packages or distributions:
>> >>   -bdist  -- build a binary distribution of Sage
>> >>   -sdist  -- build a source distribution of Sage
>> >> salvus@compute7-us:/projects/sage/sage-develop$ ls
>> >> aclocal.m4  bootstrap  config  config.status  configure.ac
>> >> local  m4README.md  src   VERSION.txt
>> >> autom4te.cache  build  config.log  configure  COPYING.txt
>> >> logs   Makefile  sage   upstream
>> >>
>> >> ---
>> >>
>> >> Also, even if bdist worked, I don't understand the docs.  What's
>> >> ??!?? Where does the bdist actually go??   It used to go in a
>> >> directory SAGE_ROOT/dist, I think.
>> >>
>> >> William
>> >>
>> >>
>> >> On Fri, May 6, 2016 at 2:42 AM, Eric Gourgoulhon 
>> >> wrote:
>> >> > As it sta

Re: [sage-devel] Re: Building binaries...

2016-05-07 Thread Volker Braun
On Saturday, May 7, 2016 at 8:51:40 PM UTC+2, William wrote:
>
> My use case is building Sage on SageMathCloud (Ubuntu 15.10 right now) 
> for people to develop on SageMathCloud (on the exact same machine).


Then just put sage in the same path for everyone (with a private union 
mount), thats really the best solution in that scenario. Getting this 
really strong deja vu right now ;-)

However, why did you have to completely eliminate all the very useful 
> functionality (related to bdist and copy) that does work fine in some 
> cases? 
>

Setting LD_LIBRARY_PATH really doesn't work for many, so we must not use it.

You can set it yourself in the environment after moving Sage, this may work 
with the aforementioned caveats about conflicts with system libraries. And 
change SAGE_LOCAL/lib/sage-current-location.txt to avoid the relocation 
error message. But its definitely not the best solution.

In any case, I don't think saving an hour or two to serve this very special 
usage is worth special casing how we build binaries.

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-07 Thread William Stein
On Sat, May 7, 2016 at 11:32 AM, Volker Braun  wrote:
> On Saturday, May 7, 2016 at 5:46:20 PM UTC+2, William wrote:
>>
>> I read that github page, but I don't know what binary-pkg actually
>> does.
>
>
> It compiles Sage in a long directory path.
>
>>
>> For context, I used to (1) build a copy of Sage, (2) possibly
>> customize it, then (3) type
>>
>>   ./sage -bdist 
>>
>> to make a binary from it.
>
>
> And then (4) the bundled gmp/readline/... would segfault your system
> gcc/bash/.., making it totally unusable. Unless you were lucky enough to use
> one of the most popular distros, where we would work around that by picking
> the right versions of libraries to not conflict.
>

My use case is building Sage on SageMathCloud (Ubuntu 15.10 right now)
for people to develop on SageMathCloud (on the exact same machine).
Obviously this does work -- we used Sage this way for 10 years very
productively.

> Then Apple disabled the DYLD_LIBRARY_PATH hack, so OSX became overnight one
> of the affected platforms where the binaries did not work.

What you did fixing that problem and doing things right using rpath is
fantastic and I greatly appreciate it!!!

However, why did you have to completely eliminate all the very useful
functionality (related to bdist and copy) that does work fine in some
cases?

 -- William



-- 
William (http://wstein.org)

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


[sage-devel] Re: Building binaries...

2016-05-07 Thread Volker Braun
On Saturday, May 7, 2016 at 5:46:20 PM UTC+2, William wrote:
>
> I read that github page, but I don't know what binary-pkg actually 
> does.  


It compiles Sage in a long directory path.
 

> For context, I used to (1) build a copy of Sage, (2) possibly 
> customize it, then (3) type 
>
>   ./sage -bdist  
>
> to make a binary from it. 
>

And then (4) the bundled gmp/readline/... would segfault your system 
gcc/bash/.., making it totally unusable. Unless you were lucky enough to 
use one of the most popular distros, where we would work around that by 
picking the right versions of libraries to not conflict.

Then Apple disabled the DYLD_LIBRARY_PATH hack, so OSX became overnight one 
of the affected platforms where the binaries did not work.

 

>
> What does this binary-pkg program do?   Does it package up an existing 
> Sage install?  If so, how do you specify which sage install it 
> packages?   Yes, I read over the yaml file but couldn't figure it out. 
>

You edit the 

repository: https://github.com/sagemath/sage.git
branch: develop

to point to the repo/branch that you want to build
 

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.


Re: [sage-devel] Re: Building binaries...

2016-05-07 Thread William Stein
On Sat, May 7, 2016 at 9:02 AM, Dima Pasechnik  wrote:
> Yes, I know, I was puzzled by this some time ago too. The reason for all
> this mess is that in order to make
> a relocatable Sage binary (more precisely, a bunch of relocateable libs etc)
> that uses rpath, it has to be built at a location that will allow
> pattern-matching on binaries to work (so it has to be some long and
> convoluted string).
> Thus your usual directory where you build Sage won't work.

Thank you for the explanation.   I wish that instead of breaking all
the existing stuff, we had just made the new parts and left the old as
an option.  (I realize the proper rpath support is useful for OS X,
etc., but it is a pain.)  Definitely one  very useful aspect of Sage,
which I had worked very hard to make, is completely gone now.

 -- Wiliam

>
> Dima
>
> On Saturday, May 7, 2016 at 4:46:20 PM UTC+1, William wrote:
>>
>> On Sat, May 7, 2016 at 2:00 AM, Volker Braun  wrote:
>> > The sage script in your PATH is outdated:
>> >
>> > $ ./sage -advanced | grep bdist
>> > $ egrep -r bdist src/
>> > $
>> >
>> > The way to build binary (relocatable) packages is
>> > https://github.com/sagemath/binary-pkg, which says:
>> >
>> > git clone https://github.com/sagemath/binary-pkg.git
>> > cd binary-pkg
>> > make bdist-sage-linux # If you are on Linux
>> > make bdist-sage-osx   # If you are on OSX
>> > ls dist/  # Built binaries will be in this directory
>>
>> I read that github page, but I don't know what binary-pkg actually
>> does.  The docs don't answer my questions.
>>
>> For context, I used to (1) build a copy of Sage, (2) possibly
>> customize it, then (3) type
>>
>>   ./sage -bdist 
>>
>> to make a binary from it.
>>
>> What does this binary-pkg program do?   Does it package up an existing
>> Sage install?  If so, how do you specify which sage install it
>> packages?   Yes, I read over the yaml file but couldn't figure it out.
>>
>>
>> 
>> I am sad that I can't copy existing sage installs, and I'm now even
>> sadder that I can't type "./sage -bdist" anymore.  Two of the most
>> important things for making sage dev easy for people are gone.   If I
>> had the time, I would make a fork of Sage that restored exactly this
>> behavior then always merge it into sage...
>> 
>>
>> William
>>
>>
>> >
>> >
>> >
>> > On Saturday, May 7, 2016 at 2:05:00 AM UTC+2, William Stein wrote:
>> >>
>> >> Either my build is completely hosed, or "sage -bdist" is completely
>> >> broken:
>> >>
>> >> salvus@compute7-us:/projects/sage/sage-develop$ time ./sage -bdist tmp
>> >> sage-run received unknown option: -bdist
>> >> usage: sage [options]
>> >> Try 'sage -h' for more information.
>> >> salvus@compute7-us:/projects/sage/sage-develop$ sage -advanced|grep
>> >> dist
>> >> Making Sage packages or distributions:
>> >>   -bdist  -- build a binary distribution of Sage
>> >>   -sdist  -- build a source distribution of Sage
>> >> salvus@compute7-us:/projects/sage/sage-develop$ ls
>> >> aclocal.m4  bootstrap  config  config.status  configure.ac
>> >> local  m4README.md  src   VERSION.txt
>> >> autom4te.cache  build  config.log  configure  COPYING.txt
>> >> logs   Makefile  sage   upstream
>> >>
>> >> ---
>> >>
>> >> Also, even if bdist worked, I don't understand the docs.  What's
>> >> ??!?? Where does the bdist actually go??   It used to go in a
>> >> directory SAGE_ROOT/dist, I think.
>> >>
>> >> William
>> >>
>> >>
>> >> On Fri, May 6, 2016 at 2:42 AM, Eric Gourgoulhon 
>> >> wrote:
>> >> > As it stands, symbolic calculus is quite severely broken in Sage 7.2:
>> >> > making
>> >> > any assumption on a symbolic variable turns it into an integer. For
>> >> > instance
>> >> > assume(x>0) yields sin(pi*x)=0. This has many undesirable
>> >> > consequences
>> >> > and
>> >> > users will probably complain a lot. Fortunately a fix is provided by
>> >> > the
>> >> > upgrade to the latest version of pynac, which is proposed in #20475.
>> >> > I
>> >> > am
>> >> > reviewing this ticket and IMHO it should be merged before releasing
>> >> > 7.2.
>> >> >
>> >> > --
>> >> > You received this message because you are subscribed to the Google
>> >> > Groups
>> >> > "sage-release" group.
>> >> > To unsubscribe from this group and stop receiving emails from it,
>> >> > send
>> >> > an
>> >> > email to sage-release...@googlegroups.com.
>> >> > To post to this group, send email to sage-r...@googlegroups.com.
>> >> > Visit this group at https://groups.google.com/group/sage-release.
>> >> > For more options, visit https://groups.google.com/d/optout.
>> >>
>> >>
>> >>
>> >> --
>> >> William (http://wstein.org)
>> >
>> > --
>> > You received this message because you are subscribed to the Google
>> > Groups
>> > "sage-release" group.
>> > To unsubscribe from this group and stop receiving emails from it, send
>> > an
>> > email to sage-release...@googlegroups.com.
>> > To post to this group, send email to sage-r...@googlegroups.co

[sage-devel] Re: Building binaries...

2016-05-07 Thread Dima Pasechnik
Yes, I know, I was puzzled by this some time ago too. The reason for all 
this mess is that in order to make
a relocatable Sage binary (more precisely, a bunch of relocateable libs 
etc) that uses rpath, it has to be built at a location that will allow 
pattern-matching on binaries to work (so it has to be some long and 
convoluted string). 
Thus your usual directory where you build Sage won't work.

Dima

On Saturday, May 7, 2016 at 4:46:20 PM UTC+1, William wrote:
>
> On Sat, May 7, 2016 at 2:00 AM, Volker Braun  > wrote: 
> > The sage script in your PATH is outdated: 
> > 
> > $ ./sage -advanced | grep bdist 
> > $ egrep -r bdist src/ 
> > $ 
> > 
> > The way to build binary (relocatable) packages is 
> > https://github.com/sagemath/binary-pkg, which says: 
> > 
> > git clone https://github.com/sagemath/binary-pkg.git 
> > cd binary-pkg 
> > make bdist-sage-linux # If you are on Linux 
> > make bdist-sage-osx   # If you are on OSX 
> > ls dist/  # Built binaries will be in this directory 
>
> I read that github page, but I don't know what binary-pkg actually 
> does.  The docs don't answer my questions. 
>
> For context, I used to (1) build a copy of Sage, (2) possibly 
> customize it, then (3) type 
>
>   ./sage -bdist  
>
> to make a binary from it. 
>
> What does this binary-pkg program do?   Does it package up an existing 
> Sage install?  If so, how do you specify which sage install it 
> packages?   Yes, I read over the yaml file but couldn't figure it out. 
>
>
>  
> I am sad that I can't copy existing sage installs, and I'm now even 
> sadder that I can't type "./sage -bdist" anymore.  Two of the most 
> important things for making sage dev easy for people are gone.   If I 
> had the time, I would make a fork of Sage that restored exactly this 
> behavior then always merge it into sage... 
>  
>
> William 
>
>
> > 
> > 
> > 
> > On Saturday, May 7, 2016 at 2:05:00 AM UTC+2, William Stein wrote: 
> >> 
> >> Either my build is completely hosed, or "sage -bdist" is completely 
> >> broken: 
> >> 
> >> salvus@compute7-us:/projects/sage/sage-develop$ time ./sage -bdist tmp 
> >> sage-run received unknown option: -bdist 
> >> usage: sage [options] 
> >> Try 'sage -h' for more information. 
> >> salvus@compute7-us:/projects/sage/sage-develop$ sage -advanced|grep 
> dist 
> >> Making Sage packages or distributions: 
> >>   -bdist  -- build a binary distribution of Sage 
> >>   -sdist  -- build a source distribution of Sage 
> >> salvus@compute7-us:/projects/sage/sage-develop$ ls 
> >> aclocal.m4  bootstrap  config  config.status  configure.ac 
> >> local  m4README.md  src   VERSION.txt 
> >> autom4te.cache  build  config.log  configure  COPYING.txt 
> >> logs   Makefile  sage   upstream 
> >> 
> >> --- 
> >> 
> >> Also, even if bdist worked, I don't understand the docs.  What's 
> >> ??!?? Where does the bdist actually go??   It used to go in a 
> >> directory SAGE_ROOT/dist, I think. 
> >> 
> >> William 
> >> 
> >> 
> >> On Fri, May 6, 2016 at 2:42 AM, Eric Gourgoulhon  
> >> wrote: 
> >> > As it stands, symbolic calculus is quite severely broken in Sage 7.2: 
> >> > making 
> >> > any assumption on a symbolic variable turns it into an integer. For 
> >> > instance 
> >> > assume(x>0) yields sin(pi*x)=0. This has many undesirable 
> consequences 
> >> > and 
> >> > users will probably complain a lot. Fortunately a fix is provided by 
> the 
> >> > upgrade to the latest version of pynac, which is proposed in #20475. 
> I 
> >> > am 
> >> > reviewing this ticket and IMHO it should be merged before releasing 
> 7.2. 
> >> > 
> >> > -- 
> >> > You received this message because you are subscribed to the Google 
> >> > Groups 
> >> > "sage-release" group. 
> >> > To unsubscribe from this group and stop receiving emails from it, 
> send 
> >> > an 
> >> > email to sage-release...@googlegroups.com. 
> >> > To post to this group, send email to sage-r...@googlegroups.com. 
> >> > Visit this group at https://groups.google.com/group/sage-release. 
> >> > For more options, visit https://groups.google.com/d/optout. 
> >> 
> >> 
> >> 
> >> -- 
> >> William (http://wstein.org) 
> > 
> > -- 
> > You received this message because you are subscribed to the Google 
> Groups 
> > "sage-release" group. 
> > To unsubscribe from this group and stop receiving emails from it, send 
> an 
> > email to sage-release...@googlegroups.com . 
> > To post to this group, send email to sage-r...@googlegroups.com 
> . 
> > Visit this group at https://groups.google.com/group/sage-release. 
> > For more options, visit https://groups.google.com/d/optout. 
>
>
>
> -- 
> William (http://wstein.org) 
>

-- 
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+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@goo