Need Cygwin testers and/or reviewers

[kcrisman]

Thanks to some great patch ideas from Mike Hansen and Dima Pasechnik,
and then a lot of boring spkg making and testing by me, we now have
spkgs that resolve all but one of the avoidable build issues for
Cygwin.

So we need testers! If you have access to a Windows box and have the
time to download Cygwin, install a few of its packages, then take a
new (say, 4.7.1.alpha3) tarball, move in the spkgs in question, and do
"make", we would really appreciate it.

Please see http://trac.sagemath.org/sage_trac/ticket/6743 and
http://trac.sagemath.org/sage_trac/wiki/CygwinPort for more
information on exactly what to do.

The tickets that still need positive review are
http://trac.sagemath.org/sage_trac/ticket/11245
http://trac.sagemath.org/sage_trac/ticket/11119
http://trac.sagemath.org/sage_trac/ticket/11504
http://trac.sagemath.org/sage_trac/ticket/11497
http://trac.sagemath.org/sage_trac/ticket/11499
That's less than half! And all these changes are very small.

Also, given that #11130 may not have positive review soon, it would be
helpful to have
http://trac.sagemath.org/sage_trac/ticket/10240
merged, since it does have positive review.

I cannot say your build will go 100% by hand, for three reasons:

1) Maxima may or may not build. It's somewhat random. Juanjo has
made some changes in ECL (since we weren't the only ones experiencing
this) that should fix this, but upstream isn't at a stable release
yet. See http://sourceforge.net/mailarchive/message.php?msg_id=27670040
If this happens, just "touch spkg/installed/maxima-5.23.2.p0" and
continue with "make" again.

2) There is a Sage library patch that needs to be added. See
http://ask.sagemath.org/question/620/after-debugging-problem-in-building-sage-library
for my dumb question about how to get the Sage install to keep going
when this happens. (Once this gets into the Sage library, this won't
be a problem.)

3) Rebasing. This is always a problem on Cygwin. But William has
some very precise instructions that will work as long as you follow
the rules for using ash on the wiki page. With luck, your computer
won't do any autoupdates while you are building and this won't happen
to you.

And Sage won't actually start, just in case anyone was wondering -
there is a segfault in Pari, see #11551. But that's a bigger goal.

Which means this is is pretty good. If I can do it, anyone can. And
it would be really good to honor the work of esp. Mike H. over many
months by getting these things in and then starting to ensure that we
don't introduce new problems! So please help.

Thanks,
- kcrisman

[RegB]

OK, I am doing it.
My initial response (as a naive Windoze user) is Gee, command line
stuff !!!
I have to learn UNIX commands to install all these packages ?

Thanks,
Where do you want feedback sent/posted ?

Best,

Reg

On Jul 1, 1:08 pm, kcrisman <kcris...@gmail.com> wrote:
> Thanks to some great patch ideas from Mike Hansen and Dima Pasechnik,
> and then a lot of boring spkg making and testing by me, we now have
> spkgs that resolve all but one of the avoidable build issues for
> Cygwin.
>
> So we need testers!  If you have access to a Windows box and have the
> time to download Cygwin, install a few of its packages, then take a
> new (say, 4.7.1.alpha3) tarball, move in the spkgs in question, and do
> "make", we would really appreciate it.
>

> Please seehttp://trac.sagemath.org/sage_trac/ticket/6743andhttp://trac.sagemath.org/sage_trac/wiki/CygwinPortfor more

> information on exactly what to do.
>

> The tickets that still need positive review arehttp://trac.sagemath.org/sage_trac/ticket/11245http://trac.sagemath.org/sage_trac/ticket/11119http://trac.sagemath.org/sage_trac/ticket/11504http://trac.sagemath.org/sage_trac/ticket/11497http://trac.sagemath.org/sage_trac/ticket/11499

> That's less than half!  And all these changes are very small.
>
> Also, given that #11130 may not have positive review soon, it would be

> helpful to havehttp://trac.sagemath.org/sage_trac/ticket/10240

> merged, since it does have positive review.
>
> I cannot say your build will go 100% by hand, for three reasons:
>
> 1) Maxima may or may not build.  It's somewhat random.  Juanjo has
> made some changes in ECL (since we weren't the only ones experiencing
> this) that should fix this, but upstream isn't at a stable release

> yet. Seehttp://sourceforge.net/mailarchive/message.php?msg_id=27670040

> If this happens, just "touch spkg/installed/maxima-5.23.2.p0" and
> continue with "make" again.
>

> 2) There is a Sage library patch that needs to be added.   Seehttp://ask.sagemath.org/question/620/after-debugging-problem-in-build...

[kcrisman]

On Jul 1, 8:37 pm, RegB <2regburg...@earthlink.net> wrote:
> OK, I am doing it.
> My initial response (as a naive Windoze user) is Gee, command line
> stuff !!!
> I have to learn UNIX commands to install all these packages ?
>
> Thanks,
> Where do you want feedback sent/posted ?

For now, right here is fine.

Yes, unfortunately for now you do need to learn some basic command
line stuff. (It's either possible now, or will be soon, to do it
somewhat more easily in one of the Mac variants.) But this is
testing something pretty subtle, so for now it is pretty hairy. Good
luck!

- kcrisman

[Sazzad]

I have to learn UNIX commands to install all these packages ?

Best,
Reg

Not that much. But, you have to read and hack a lot to understand the total process. To install cygwin you just need to download, the around 500K installer. Then you either select everything to download and install or select only those that are needed. This is unfortunately difficult for the beginners and even sometimes for the advanced users. If your internet bandwidth is good and if the download size isn't a big issue, then I would suggest you to download everything. Cygwin is an amazing environment. You have almost every compilers, java, c, c++, python, perl, clisp etc, in cygwin.

Once you download and install, Cygwin you can simply download the sage spkg and read the readme.txt or install.txt files and specially the documentations for the developers. I've found the documentation really helpful.

Hope it helps in some way.

I'll now try to download the mentioned sage tarball. Let see.

--
Sincerely,
Sazzad Bin Kamal

[RegB]

For MS-Windows users with little/no expectations;
To install ALL of Cygwin requires about a 1.36 Gig download.
This is just a note to help set an expectation, don't let it dissuade
anyone (-:.
The setup program is sufficiently "pointy/clicky" that no *nix
knowledge is required, just
an awareness that the branches (& twigs) of the tree representing
optional packages are to be clicked
on to rotate between Default->Install->Reinstall->Uninstall

So far I have;
1) Performed an initial bare bones (default) install, which did not
include Cygwin/X
2) Tried to add all of the X11 packages, that resulted in an inability
to start the X server.
3) Figured that although the setup program APPEARS TO resolve
dependencies there can still be some things missing, so I will do only
FULL installs of EVERYTHING from now on.
4) Kept every suspect activity restrained to my virus detector's sand
box (Avast).
I suspect that this will cause some problems fairly soon.
5) My last full install gave the following messages;
Package: libglade2.0_0
libglade2.0.sh exit code 2
Package: exim
exim.sh exit code 1
With a suggestion to check /var/log/setup.log.full
but that file is from the original bare bones install, i.e. was not
overwritten or appended to.

Sorry I have nothing to report on the INTENDED experiment yet,
hopefully these notes will help
other struggling newbies through the first steps.

[Dima Pasechnik]

http://cygwin.com/faq/faq.using.html#faq.using.bloda

> 5) My last full install gave the following messages;
> Package: libglade2.0_0
>         libglade2.0.sh exit code 2
> Package: exim
>         exim.sh exit code 1
> With a suggestion to check /var/log/setup.log.full
> but that file is from the original bare bones install, i.e. was not
> overwritten or appended to.

you won't need either of these two things, so just ignore these
warnings.

[Sazzad]

This is the error message I've got:

./spkg-install: line 248: /cygdrive/e/Sagemath_4_7/sage-4.7/local/bin/python: No such file or directory math module failed to import

sage: An error occurred while installing python-2.6.4.p10

OS : Windows XP SP3 + Cygwin

Processor : 1.8 GHz Intel Pentium 4 processor

[kcrisman]

On Jul 4, 12:25 pm, Sazzad <sazzadbinka...@gmail.com> wrote:
> This is the error message I've got:
>

> *./spkg-install: line 248:

> /cygdrive/e/Sagemath_4_7/sage-4.7/local/bin/python: No such file or

> directory math module failed to import*

>
> sage: An error occurred while installing python-2.6.4.p10
>
> OS : Windows XP SP3 + Cygwin
> Processor : 1.8 GHz Intel Pentium 4 processor

Thanks, Sazzad. Is it possible to give a little more of the error
message? If there is anything about stack or dump or unable to access
or something, it's likely one would have to "rebase". See the
instructions at http://trac.sagemath.org/sage_trac/wiki/CygwinPort
about "rebasing" - it was a little intimidating at first, but
eventually I figured out how to do it. Then you should be able to do
"make" again to continue the build.

[RegB]

Newbie notes continued;
The Avast sand box DID INDEED give me problems, so I have done total
uninstall, deleted the C:\cygwin folder, install ALL again allowing
normal installs
when Avast complained (twice, maybe 3 times) towards the end of the
install process.
This is NOT something I had to sit and watch, it is at about the 98%
or 99% done point.

Again, for newbies;
wget is the command for fetching packages from URLs, i.e. http://
addresses.
I haven't (yet) figured out how to copy a link and paste it into a
bash shell immediately
after typing "wget ", so I have been hovering my mouse over the links
to display them in
Firefox on the MSwindows side and typing the full address to the wget
command in the cygwin terminal window.
Crude ? maybe, but for now I'm OK with that.
I could not find a sage-4.7.1-alpha3.tar file, so am downloading an
alpha4, also with wget.

Next;
Figure out the patching, presumably this is editing ? I vaguely
remember ed.
Unpack the tar files to the "right" place(s).
DO the experimental make.
Report back.

Happy 4th July to all who celebrate this day.

On Jul 4, 1:55 pm, kcrisman <kcris...@gmail.com> wrote:
> On Jul 4, 12:25 pm, Sazzad <sazzadbinka...@gmail.com> wrote:
>
> > This is the error message I've got:
>
> > *./spkg-install: line 248:
> > /cygdrive/e/Sagemath_4_7/sage-4.7/local/bin/python: No such file or
> > directory math module failed to import*
>
> > sage: An error occurred while installing python-2.6.4.p10
>
> > OS : Windows XP SP3 + Cygwin
> > Processor : 1.8 GHz Intel Pentium 4 processor
>
> Thanks, Sazzad.   Is it possible to give a little more of the error
> message?  If there is anything about stack or dump or unable to access
> or something, it's likely one would have to "rebase".    See the

> instructions athttp://trac.sagemath.org/sage_trac/wiki/CygwinPort

[Dr. David Kirkby]

On 07/ 2/11 04:05 PM, Sazzad wrote:
>>
>> I have to learn UNIX commands to install all these packages ?
>
> Best,
>> Reg
>
>
> Not that much. But, you have to read and hack a lot to understand the total
> process. To install cygwin you just need to download, the around 500K
> installer. Then you either select everything to download and install or
> select only those that are needed. This is unfortunately difficult for the
> beginners and even sometimes for the advanced users. If your internet
> bandwidth is good and if the download size isn't a big issue, then I would
> suggest you to download everything. Cygwin is an amazing environment. You
> have almost every compilers, java, c, c++, python, perl, clisp etc, in
> cygwin.

I would have thought if you download everything you have a fairly high
probability of introducing an incompatibility between two or more items.

This is an easy approach, but I'm not convinced it is a particularly sound
approach.

--
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

[kcrisman]

On Jul 4, 4:50 pm, RegB <2regburg...@earthlink.net> wrote:
> Newbie notes continued;
> The Avast sand box DID INDEED give me problems, so I have done total
> uninstall, deleted the C:\cygwin folder, install ALL again allowing
> normal installs
> when Avast complained (twice, maybe 3 times) towards the end of the
> install process.
> This is NOT something I had to sit and watch, it is at about the 98%
> or 99% done point.

Yeah, I would actually do as minimal an install of Cygwin as possible
for the reasons David Kirkby mentions. I think I found all the
various needed things (perl, for instance) in my own testing, and put
them on that wiki page.

> Again, for newbies;
> wget is the command for fetching packages from URLs, i.e. http://
> addresses.

Yes! And sadly, I have no better way to do this than you do. It's
also possible to right-click on an address on the webpage to copy,
then to hover over the title bar of the Cygwin shell window, right-
click and find the edit->paste thing. Both are annoying, you are
right.

> Crude ? maybe, but for now I'm OK with that.
> I could not find a sage-4.7.1-alpha3.tar file, so am downloading an
> alpha4, also with wget.

Well, and alpha4 is now the current dev release.

>
> Next;
> Figure out the patching, presumably this is editing ?  I vaguely
> remember ed.

You should not have to do anything here until well into the build.
When you get an error "building sage.rings.factorint" extension or
something many hours in, you will have one very small edit to make,
which is documented. You can just use any editor you want (I like
nano/pico because the instructions are at the bottom, but you may have
to install nano on Cygwin) to make the change on that ticket, in sage-
x.y.z/devel/sage/sage/rings/factorint.pyx.

> Unpack the tar files to the "right" place(s).

Yes, unpack the alpha Sage, and then take all these .spkg files and
move them to sage-x.y.z/spkg/standard and Sage should do the rest.

> DO the experimental make.
> Report back.

Please! That would be great.

> Happy 4th July to all who celebrate this day.

Indeed.

[Sazzad]

Thanks for your instructions.

On Mon, Jul 4, 2011 at 11:55 PM, kcrisman <kcri...@gmail.com> wrote:

On Jul 4, 12:25 pm, Sazzad <sazzadbinka...@gmail.com> wrote:
> This is the error message I've got:
>
> *./spkg-install: line 248:
> /cygdrive/e/Sagemath_4_7/sage-4.7/local/bin/python: No such file or
> directory math module failed to import*
>
> sage: An error occurred while installing python-2.6.4.p10
>
> OS : Windows XP SP3 + Cygwin
> Processor : 1.8 GHz Intel Pentium 4 processor

Thanks, Sazzad.   Is it possible to give a little more of the error
message?

I'm at office now. I'll upload the install log file once I go back to home at the evening.

If there is anything about stack or dump or unable to access
or something,

So far I remember, there was no such problem, but I'll reexamine it today.

 

it's likely one would have to "rebase".    See the
instructions at http://trac.sagemath.org/sage_trac/wiki/CygwinPort
about "rebasing" - it was a little intimidating at first, but
eventually I figured out how to do it.  Then you should be able to do
"make" again to continue the build.

Ok, I'll give it a try and let you know the updates.

Stay fine. 

[RegB]

I messed up.
I edited the sage file to set SAGE_ROOT
I did export SAGE_PORT='yes'
UNFORTUNATELY I copied the listed spkg files to the top level
directory /usr/local/sage-4.7.1.alpha4
I should have put them in /usr/local/sage-4.7.1.alpha4/spkg/standard

It seems to be running OK, but I suspect the result will be
meaningless ?

I will probably let it run tonight, just to see if it crashes out for
other reasons.
I will try another run tomorrow with the packages in the right place.

Note for other Newbies;
The emacs is VERY EASY to use, to open it right clock on the drag bar
of the Cygwin bash shell window.

WRT installing ALL;
There seems to be dependency checking, including checking for and semi-
automatic
removal of obsolete packages.
It is my belief that dependency checks (should) include checks for
absence of the undesirable
as much as they check for presence of the desirable.
Of course there is always the possibility that a package needed by
whatever one is trying to do
could be considered undesirable, but I only got error messages on;
libglade2.0_0 and exim so I think I may be OK ?

[RegB]

I also got an error on installing python-2.6.4.p10
The end of the log is as follows;
===========================================
creating build/temp.cygwin-1.7.9-i686-2.6/usr/local/sage-4.7.1.alpha4/
spkg/build/python-2.6.4.p10/src
creating build/temp.cygwin-1.7.9-i686-2.6/usr/local/sage-4.7.1.alpha4/
spkg/build/python-2.6.4.p10/src/Modules
gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -O3 -Wall -Wstrict-
prototypes -I. -I/usr/local/sage-4.7.1.alpha4/spkg/build/
python-2.6.4.p10/src/./Include -I. -IInclude -I./Include -I/usr/local/
sage-4.7.1.alpha4/local/include -I/usr/local/sage-4.7.1.alpha4/spkg/
build/python-2.6.4.p10/src/Include -I/usr/local/sage-4.7.1.alpha4/spkg/
build/python-2.6.4.p10/src -c /usr/local/sage-4.7.1.alpha4/spkg/build/
python-2.6.4.p10/src/Modules/_struct.c -o build/temp.cygwin-1.7.9-
i686-2.6/usr/local/sage-4.7.1.alpha4/spkg/build/python-2.6.4.p10/src/
Modules/_struct.o
5 [main] python 6044 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
9709 [main] python 6044 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
5 [main] python 5640 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
3097 [main] python 5640 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
6 [main] python 3912 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
4137 [main] python 3912 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
6 [main] python 6004 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
3511 [main] python 6004 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
5 [main] python 2072 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
2695 [main] python 2072 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
5 [main] python 2184 exception::handle: Exception:
STATUS_ACCESS_VIOLATION
2834 [main] python 2184 open_stackdumpfile: Dumping stack trace to
python.exe.stackdump
6 [main] python 4928 fork: child -1 - died waiting for longjmp
before initialization, retry 0, exit code 0x600, errno 11
error: Resource temporarily unavailable
make[2]: *** [sharedmods] Error 1
make[2]: Leaving directory `/usr/local/sage-4.7.1.alpha4/spkg/build/
python-2.6.4.p10/src'
Error building Python.

real 19m25.122s
user 4m56.365s
sys 9m50.157s

sage: An error occurred while installing python-2.6.4.p10

================================================
then the "please send logs to..." lines.

Not sure if this is from the same cause ?
I happen to be running on Vista.

On Jul 4, 12:25 pm, Sazzad <sazzadbinka...@gmail.com> wrote:

> This is the error message I've got:
>

> *./spkg-install: line 248:

> /cygdrive/e/Sagemath_4_7/sage-4.7/local/bin/python: No such file or

> directory math module failed to import*

[Sazzad]

I ran grep "error" on the install.log and it seems the problem is with pycurses.

Here are some lines:

Modules/symtablemodule.o Modules/xxsubtype.o \
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/./Include/py_curses.h:50:20: error: curses.h: No such file or directory
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/./Include/py_curses.h:78: error: expected specifier-qualifier-list before ‘WINDOW’
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cursesmodule.c:141: error: ‘FALSE’ undeclared here (not in a function)
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cursesmodule.c:183: error: ‘ERR’ undeclared (first use in this function)
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cursesmodule.c:183: error: (Each undeclared identifier is reported only once
/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cursesmodule.c:183: error: for each function it appears in.)

 

.....................

...... The road goes on and on :D

.....................

And, lastly,

 

/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_curses_panel.c:411: error: ‘PANEL’ undeclared (first use in this function) 

/cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_curses_panel.c:411: error: ‘pan’ undeclared (first use in this function) 

sage: An error occurred while installing python-2.6.4.p10 

correctly and load a subshell for you to debug the error:

By the way, I did not get STATUS_ACCESS_VIOLATION error. But, I certainly got this error:

gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -O3 -Wall -Wstrict-
prototypes -I. -I/usr/local/sage-4.7.1.alpha4/spkg/build/
python-2.6.4.p10/src/./Include -I. -IInclude -I./Include -I/usr/local/
sage-4.7.1.alpha4/local/include -I/usr/local/sage-4.7.1.alpha4/spkg/
build/python-2.6.4.p10/src/Include -I/usr/local/sage-4.7.1.alpha4/spkg/
build/python-2.6.4.p10/src -c /usr/local/sage-4.7.1.alpha4/spkg/build/
python-2.6.4.p10/src/Modules/_struct.c -o build/temp.cygwin-1.7.9-
i686-2.6/usr/local/sage-4.7.1.alpha4/spkg/build/python-2.6.4.p10/src/
Modules/_struct.o

[kcrisman]

@RegB:

> I edited the sage file to set SAGE_ROOT

Yeah, don't edit anything.

> I did export SAGE_PORT='yes'

Good.

> UNFORTUNATELY I copied the listed spkg files to the top level
> directory /usr/local/sage-4.7.1.alpha4
> I should have put them in /usr/local/sage-4.7.1.alpha4/spkg/standard
>
> It seems to be running OK, but I suspect the result will be
> meaningless ?

Not meaningless, but it will stop building at the ones which aren't
upgraded, most likely. I'd untar again and start over.

>       5 [main] python 2184 exception::handle: Exception:
> STATUS_ACCESS_VIOLATION
>    2834 [main] python 2184 open_stackdumpfile: Dumping stack trace to
> python.exe.stackdump
>       6 [main] python 4928 fork: child -1 - died waiting for longjmp
> before initialization, retry 0, exit code 0x600, errno 11
> error: Resource temporarily unavailable
> make[2]: *** [sharedmods] Error 1
> make[2]: Leaving directory `/usr/local/sage-4.7.1.alpha4/spkg/build/
> python-2.6.4.p10/src'
> Error building Python.

That is the rebase problem, on the other hand. See the CygwinPort
wiki page and/or just doing rebaseall in ash in the Windows command
line (again, see http://trac.sagemath.org/sage_trac/wiki/CygwinPort
for more details and links to even more details).

@Sazzad:


> > /cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cu rsesmodule.c:183:

> > error: for each function it appears in.)

I've seen that too. For some reason rebasing seems to allow me to
continue with that as well. I don't know that the curses actually
builds, but at any rate it keeps going. See Comment 5 on
http://trac.sagemath.org/sage_trac/wiki/CygwinPort at the Windows 7
Python bullet point.

[Dima Pasechnik]

On Jul 6, 3:37 pm, kcrisman <kcris...@gmail.com> wrote:
> @RegB:
>
> > I edited the sage file to set SAGE_ROOT
>
> Yeah, don't edit anything.
>
> > I did export SAGE_PORT='yes'
>
> Good.
>
> > UNFORTUNATELY I copied the listed spkg files to the top level
> > directory /usr/local/sage-4.7.1.alpha4
> > I should have put them in /usr/local/sage-4.7.1.alpha4/spkg/standard
>
> > It seems to be running OK, but I suspect the result will be
> > meaningless ?

well, open a Cygwin terminal, cd to /usr/local/sage-4.7.1.alpha4 and
then do
mv *.spkg spkg/standard/

(or perform the equivalent of this in Widows Exploder...)

the latter, /usr/local/sage-4.7.1.alpha4/spkg/standard/
(which Widows would translate as C:\cygwin\usr\local
\sage-4.7.1.alpha4\spkg\standard, assuming you
installed Cygwin in the usual place) is the location where all the
necessary spkg files should be.

>
> Not meaningless, but it will stop building at the ones which aren't
> upgraded, most likely.    I'd untar again and start over.
>
> >       5 [main] python 2184 exception::handle: Exception:
> > STATUS_ACCESS_VIOLATION
> >    2834 [main] python 2184 open_stackdumpfile: Dumping stack trace to
> > python.exe.stackdump
> >       6 [main] python 4928 fork: child -1 - died waiting for longjmp
> > before initialization, retry 0, exit code 0x600, errno 11
> > error: Resource temporarily unavailable
> > make[2]: *** [sharedmods] Error 1
> > make[2]: Leaving directory `/usr/local/sage-4.7.1.alpha4/spkg/build/
> > python-2.6.4.p10/src'
> > Error building Python.
>
> That is the rebase problem, on the other hand.   See the CygwinPort
> wiki page and/or just doing rebaseall in ash in the Windows command

> line (again, seehttp://trac.sagemath.org/sage_trac/wiki/CygwinPort

> for more details and links to even more details).
>
> @Sazzad:
>
> > > /cygdrive/path_to_sage/sage-4.7/spkg/build/python-2.6.4.p10/src/Modules/_cu rsesmodule.c:183:
> > > error: for each function it appears in.)
>
> I've seen that too.  For some reason rebasing seems to allow me to
> continue with that as well.  I don't know that the curses actually

> builds, but at any rate it keeps going.  See Comment 5 onhttp://trac.sagemath.org/sage_trac/wiki/CygwinPortat the Windows 7
> Python bullet point.

[RegB]

Thanks,
Since my last newbie note entry I was able to rebase the dlls
following William's directions.
I fumbled with this for a LONG time.
From an xterm rebaseall seemed to respond only when
given no parameters, but was not found when I tried to
use -T for specifying a file input.
Odd that it worked when I eventually resorted to a regular
Cygwin Bash Shell. I spelled out the path explicitly in all
cases.

It SEEMS that Windoze file explorer, Wordpad and Notepad
on the Windowze side work just fine on log files and
presumably any text files in the Cygwin file system.
So that MIGHT be a reasonable way for windows type folk
to move things around and edit them - though I am not
sure that ownership, privs, protection stuff would all perc
through correctly.
If/when I get through all this I may try downloading .spkg
files via windows side and doing drag/drop into Cygwin/Sage
folders to see if this works - maybe.


Last night it got all the way through to what I am fairly sure is
the failure that requires to factorint patch.
85 packages appear to have been installed, I haven't looked at the
logs
in any detail, though I did notice that the times stamps for Atlas
indicated a very quick install, so it probably took some defaults
right
away and did no tuning.
About 8 or 9 hours

> > builds, but at any rate it keeps going.  See Comment 5 onhttp://trac.sagemath.org/sage_trac/wiki/CygwinPortatthe Windows 7
> > Python bullet point.

[kcrisman]

> Since my last newbie note entry I was able to rebase the dlls
> following William's directions.

Yes, rebasing is annoying. But not as bad as I thought it was going
to be.

> I fumbled with this for a LONG time.
> From an xterm rebaseall seemed to respond only when
> given no parameters, but was not found when I tried to
> use -T for specifying a file input.

The "correct" way to use rebaseall is by closing all Cygwin stuff,
opening a Windows command prompt, doing "ash", and then using
rebaseall. I think I linked to full instructions for this and
William's may have mentioned this too, but if not, I will try to write
something up (unfortunately, the next couple weeks are a lousy time
for me to do this).

>
> If/when I get through all this I may try downloading .spkg
> files via windows side and doing drag/drop into Cygwin/Sage
> folders to see if this works - maybe.

It's true that downloading with wget might be annoying, but the syntax
is straightforward enough, even for me:

wget http://location_of_spkg

mv this.spkg spkg/standard

or whatever, depending on your current location.

> Last night it got all the way through to what I am fairly sure is
> the failure that requires to factorint patch.

You can just do that one manually - I think I give instructions, but
going in the sage-4.7.1.../devel/sage/module_list.py and doing the
change indicated on the relevant ticket by hand, saving it, and then
continuing to follow the instructions linked to (going to the
$SAGE_ROOT directory, doing ./sage -b, etc.) should work fine.

> 85 packages appear to have been installed, I haven't looked at the
> logs

That's great!

> away and did no tuning.

I think it just relies on the lapack we have Cygwin install.

> About 8 or 9 hours

That sounds about right. It sounds like you're just about set, then!
Good luck.

[RegB]

I was able to do the edit of module_list.py "by hand" in a very
Windows'ey way, i.e. with WordPad.
As the cygwin port moves forwards it may be worth encouraging
other test participants to use their windows side tools/expertise ?

Then the recovery from that failure worked, so I was able to move
on to the already documented failure of Maxima to build.
I merely deleted the spkg from standard to get that out of the way.
The next run produced the already documented segmentation fault.

So, all quite predictable - I doubt that I have proven anything other
than "Yeah, same here", though there may be some use to that.

Not sure if I can chase down any of the causes, I am both time and
expertise limited, so it is unlikely that I would be first to find
causes
even if I did.

On Jul 11, 1:29 pm, kcrisman <kcris...@gmail.com> wrote:
> > Since my last newbie note entry I was able to rebase the dlls
> > following William's directions.
>
> Yes, rebasing is annoying.  But not as bad as I thought it was going
> to be.
>
> > I fumbled with this for a LONG time.
> > From an xterm rebaseall seemed to respond only when
> > given no parameters, but was not found when I tried to
> > use -T for specifying a file input.
>
> The "correct" way to use rebaseall is by closing all Cygwin stuff,
> opening a Windows command prompt, doing "ash", and then using
> rebaseall.  I think I linked to full instructions for this and
> William's may have mentioned this too, but if not, I will try to write
> something up (unfortunately, the next couple weeks are a lousy time
> for me to do this).
>
>
>
> > If/when I get through all this I may try downloading .spkg
> > files via windows side and doing drag/drop into Cygwin/Sage
> > folders to see if this works - maybe.
>
> It's true that downloading with wget might be annoying, but the syntax
> is straightforward enough, even for me:
>

> wgethttp://location_of_spkg

[kcrisman]

On Jul 12, 7:40 am, RegB <2regburg...@earthlink.net> wrote:
> I was able to do the edit of module_list.py "by hand" in a very
> Windows'ey way, i.e. with WordPad.
> As the cygwin port moves forwards it may be worth encouraging
> other test participants to use their windows side tools/expertise ?

Perhaps! I don't actually know how to use those tools very well, and
in particular I don't know how to find the Cygwin stuff in the Windows
filesystem, so I wouldn't want to be the one to write this.

> Then the recovery from that failure worked, so I was able to move
> on to the already documented failure of Maxima to build.
> I merely deleted the spkg from standard to get that out of the way.
> The next run produced the already documented segmentation fault.

Excellent! That means that we now have an independent verification
that the steps in question at http://trac.sagemath.org/sage_trac/ticket/6743
work!

> So, all quite predictable - I doubt that I have proven anything other
> than "Yeah, same here", though there may be some use to that.

Yes, that is *very* useful. We need independent review for any
changes in Sage.

Dima, William, Mike - do you think this would be enough to give #6743
positive review? We'd have to change the ticket to "get Cygwin to
build automatically, except for Maxima", but Juanjo apparently has
taken care of this elsewhere. So one could close #6743, open a new
ticket for Maxima, and then we'd be much closer to happiness.

At the very least, RegB, you could give positive review (or at least
information pertaining to your success, esp. which versions of Cygwin
and Windows) on

http://trac.sagemath.org/sage_trac/ticket/11119
http://trac.sagemath.org/sage_trac/ticket/11504
http://trac.sagemath.org/sage_trac/ticket/11497

http://trac.sagemath.org/sage_trac/ticket/11245

Full positive review of these would need someone to look at the spkgs
too.

You could definitely give positive review to this one based on your
experience, the Sage library change, which would be great to get in:
http://trac.sagemath.org/sage_trac/ticket/11499

Instructions for getting an account (using your real name, though) are
at http://trac.sagemath.org/sage_trac/

> Not sure if I can chase down any of the causes, I am both time and
> expertise limited, so it is unlikely that I would be first to find
> causes
> even if I did.

Not looking for any of that! Just the testing is often enough,
because there are so few Sage devels who have access to Windows (and
time/motivation to test there).

Thank you.

- kcrisman

[Dima Pasechnik]

On Jul 9, 6:22 pm, RegB <2regburg...@earthlink.net> wrote:
> Thanks,
> Since my last newbie note entry I was able to rebase the dlls
> following William's directions.
> I fumbled with this for a LONG time.
> From an xterm rebaseall seemed to respond only when
> given no parameters, but was not found when I tried to
> use -T for specifying a file input.
> Odd that it worked when I eventually resorted to a regular
> Cygwin Bash Shell.  I spelled out the path explicitly in all
> cases.
>
> It SEEMS that Windoze file explorer, Wordpad and Notepad
> on the Windowze side work just fine on log files and
> presumably any text files in the Cygwin file system.
> So that MIGHT be a reasonable way for windows type folk
> to move things around and edit them - though I am not
> sure that ownership, privs, protection stuff would all perc
> through correctly.

indeed. Also watch out for Cygwin's links, which are not quite the
same thing as Windows shortcuts.
And of course the recent Windows insanity called "virtual files" (i.e.
"You can't change a file? No problem, Windows will create a copy for
you
and write things there...") isn't understood by Cygwin.

[Dima Pasechnik]

On Jul 12, 1:40 pm, RegB <2regburg...@earthlink.net> wrote:
> I was able to do the edit of module_list.py "by hand" in a very
> Windows'ey way, i.e. with WordPad.

this might mess things up, as by default Windows encodes line breaks
in text files differently from Cygwin (or any Unix system).
(well, there is a Cygwin command dos2unix to fix these, in case...)

You might want to use a Unix-aware text editor instead...
(my tool of choice is Vim)

[kcrisman]

Here is the current status.

1) The following just need ANYONE to look at the spkgs and make sure
they are constructed properly, have committed changes. Should not be
trouble, then they will be in the 4.7.2 series, hopefully.

http://trac.sagemath.org/sage_trac/ticket/11119
http://trac.sagemath.org/sage_trac/ticket/11504


2) This needs someone to do sanity check on the spkg, and needs
someone with Windows 7 to try a sage -f, since that is where the
problem is. Some sanity checks on other equipment wouldn't hurt, but
it shouldn't be a problem. Could still make it in the 4.7.2 series -
Dima?

http://trac.sagemath.org/sage_trac/ticket/11245

3) This needs someone to do the check on the spkg and test with XP.
Note it's only been observed on certain XP boxes, so it may be fixed
by a newer Cygwin. Slightly lower priority, as XP gets phased out,
but still useful.

http://trac.sagemath.org/sage_trac/ticket/11497

4) This needs testing and check on several Windows systems, though it
should work fine. See the ticket for explanation of why it was split
off.

http://trac.sagemath.org/sage_trac/ticket/11635

The rest have positive review!

Then we need to figure out what to do about ECL/Maxima (Juanjo will
eventually provide a fix but that may be a bit), and need someone to
figure out how to fix #11551 with the initialization of Pynac's I
causing an mpfr number being brought into Pari segfaulting. Ideas
welcome.

- kcrisman