python spkg build fails on Ubuntu 11.04

pipedream commented 13 years ago

http://groups.google.com/group/sage-devel/browse_thread/thread/593b9a4124f5075d

This patch is a fix for Ubuntu 11.04. Ubuntu already patch their python like this. It is not a python bug. It is the Ubuntu+python combination, see the above thread for an answer I got from bugs.python.org. It potentially affects future multiarch Debian/Ubuntu and derivatives. (At some point python 2.7 series may be patched upstream though.)

Note for the release manager

Just merge http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

ignore all the patches attached to the ticket, as they are all committed in the repository.

Upstream: None of the above - read trac for reasoning.

Component: build

Author: Jan Groenewald, David Kirkby

Reviewer: David Kirkby, Jan Groenewald, Dan Drake

Merged: sage-4.7.rc1

Issue created by migration from https://trac.sagemath.org/ticket/11243

pipedream commented 13 years ago

Attachment: spkg-install.gz

Attachment: SPKG.txt

pipedream commented 13 years ago

Attachment: Setup.dist.gz

pipedream commented 13 years ago

Attachment: Setup.dist.patch.gz

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Reviewer: David Kirkby

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:1

Hi Jan, Could you please put the trac number in SPKG.txt, then create a new .spkg so it can be reviewed.

It is better if you attach a Mercurial patch showing your changes - normally something like

$ hg commit
$ hg export tip > $HOME/11243-Python-Ubuntu-issue.patch

then attach that file to the trac ticket. There's no need to have all the other files attached.

You will need to put something like this in $HOME/.hgrc if you don't already have it - though obviously changing the name and email

drkirkby@hawk:~$ cat .hgrc
[ui]
username = David Kirkby <david.kirkby@onetel.net>

[extensions]
# Enable the Mercurial queue extension.
hgext.mq =

My biggest concern with this ticket is that it impacts every operating system, whereas the problem is only observed on one alpha version on one Linux distribution.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Author: Jan Groenewald

pipedream commented 13 years ago

comment:2

Thanks Dave. I was still busy reading the developer guides. I have now changed the patch to check it is running on Ubuntu 11.04, only apply if so, tested it on Ubuntu 11.04, and hg exported the attached patch. I hope I did it right.

Note I think this affects, according to http://hg.python.org/cpython/rev/bd0f73a9538e all future multiarch debian/ubuntu (and probably other derivatives of those). My immediate concern is Ubuntu 11.04, due for release in a week, but then this will be dormant for 6 months, and there may be upstream fixes in python 2.7 at that time, and sage may be using python 2.7.

pipedream commented 13 years ago

Attachment: 11243-Python-Ubuntu-issue.patch.gz

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:3

The 'lsb_release' is not portable. It is certainly not a POSIX command, and I have no idea if it will be found on all Linux distributions.

drkirkby@hawk:~$ lsb_release -d
bash: lsb_release: command not found

You could perhaps change your test to:

if [ "x$UNAME" = xLinux ] && [ "x`lsb_release -d`" = "xDescription:       Ubuntu 11.04" ] ; then

but I've no idea if that will work on all Linux systems.

dave

pipedream commented 13 years ago

comment:4

Does it matter if it fails? If the string doesn't match, no patch is applied, as it should be.

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:5

Replying to @pipedream:

Does it matter if it fails? If the string doesn't match, no patch is applied, as it should be.

Yes it does. It is unprofessional to use a command which fails like this, as it is disconcerting to those who don't have the command.

A better test is:

if [ `command -v lsb_release` ]  && [ "x`lsb_release -d`" = "xDescription:       Ubuntu 11.04" ] ; then

Then the second part will only get executed if 'lsb_release' exists. That should be safe on any platform, even Linux systems for which the command 'lsb_release' does not exist. The program 'command' will exist on any operating system conforming to the POSIX standard.

http://pubs.opengroup.org/onlinepubs/9699919799/utilities/command.html

While Linux does not conform to POSIX standards, any distribution I am aware of will have an executable called 'command'

As such, I think that would be a sensible improvement.

The next issue is that the patch copies a revised version of 'patches/Setup.dist'. While it is true the developers guide might still say to do that, it has been agreed to use 'patch'. So you should create a unified diff patch and apply that patch. Not only does that save space, but it allows more than one patch to be applied easily, whereas the use of 'cp' caused problems. You can guarantee that 'patch' will be available, as there is a Sage package for that.

Note also your Mercurial patch should have the contents of any patch you have.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:6

I've created a package which I think addresses the problem although I can't check it as I don't have a Ubuntu 11.04 system set up. This should

Use a portable method to check for Ubuntu 11.04 which will not generate any error messages on systems which do not have the 'lsb_release' command.
Builds the 'crypto' module on Ubuntu 11.04
Changes the method of checking that the 'haslib' and 'math' modules import properly by using a loop rather than repeating almost identical code.
Check that the 'crypto' module actually imports properly on Ubuntu 11.04

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Description changed:

--- 
+++ 
@@ -4,15 +4,11 @@

 If this harms building on other systems, I can make it specific to checking the Ubuntu version and only then applying the patch (using lsb_release -d).

-To test, untar SAGE_ROOT/spkg/standard/python-2.6.4.p9.spkg, in there replace with attached versions:
+'`11243-build-crypto-on-Ubuntu-11.04.patch`' is for review purposes only. 

-SPKG.txt
-spkg-install
-patches/Setup.dist
-patches/Setup.dist.patch
+An updated package, which should fix the problem can be found at:
+http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

-tar up as python 2.6.4.p10.spkg
-sage -f python-2.6.4.p10.spkg

-Revert with sage -f python-2.4.6.p9.spkg

+

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Changed author from Jan Groenewald to Jan Groenewald, David Kirkby

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:7

I noticed a couple of typos - marking needs_work until I fix them.

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Description changed:

--- 
+++ 
@@ -4,11 +4,9 @@

 If this harms building on other systems, I can make it specific to checking the Ubuntu version and only then applying the patch (using lsb_release -d).

-'`11243-build-crypto-on-Ubuntu-11.04.patch`' is for review purposes only. 
+'`11243-build-crypt-on-Ubuntu-11.04.patch`' is for review purposes only. 

 An updated package, which should fix the problem can be found at:
 http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

-
-

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:9

This should be OK now. One should see the following on all platforms:

Sleeping for three seconds before testing python
math module imported OK
hashlib module imported OK
crypt module imported OK

real    1m37.699s
user    2m7.686s
sys     0m17.449s
Successfully installed python-2.6.4.p10

Dave

pipedream commented 13 years ago

comment:10

From the patch:

-#crypt cryptmodule.c # -lcrypt # crypt(3); needs -lcrypt on some systems +crypt cryptmodule.c # -lcrypt # crypt(3); needs -lcrypt on some systems like Ubuntu 11.04

The -lcrypt needs to be uncommented as well. It is still commented in that line.

pipedream commented 13 years ago

comment:11

Beg your pardon. Better formatting.

From the patch:

-#crypt cryptmodule.c # -lcrypt # crypt(3); needs -lcrypt on some systems
+crypt cryptmodule.c # -lcrypt  # crypt(3); needs -lcrypt on some systems like Ubuntu 11.04

The -lcrypt needs to be uncommented as well. It is still commented in that line.

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:12

I've changed the patch. Can you give that a try on Ubuntu 11.04.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:13

I obviously uploaded a new package too:

http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Attachment: 11243-build-crypt-on-Ubuntu-11.04.patch.gz

Patch to force the 'crypt' module to build on Ubuntu 11.04. Check this module imports properly on all platforms. Also simplify checking that modules import OK. FOR REVIEW PURPOSES ONLY - THE CHAGES ARE IN THE .SPKG

pipedream commented 13 years ago

comment:14

The line in spkg-install

`patch -p0 < patches/Modules.Setup.dist.patch`

should not be in back substituted with backticks.

pipedream commented 13 years ago

comment:15

Wehn testing the lsb_release -d string, the space between "Description:" and "Ubuntu 11.04" must be a tab character.

pipedream commented 13 years ago

comment:16

Attachment: 11243-crypt-module-Ubuntu11.04.patch.gz

I've added those last two changes. Attached patch for testing and the SPKG with the changes is at http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Changed reviewer from David Kirkby to David Kirkby, Jan Groenewald

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:17

That looks good to me. I've set it to positive review.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Description changed:

--- 
+++ 
@@ -4,9 +4,10 @@

 If this harms building on other systems, I can make it specific to checking the Ubuntu version and only then applying the patch (using lsb_release -d).

-'`11243-build-crypt-on-Ubuntu-11.04.patch`' is for review purposes only. 
+## Note for the release manager

-An updated package, which should fix the problem can be found at:
-http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg
+Just merge 
+
+http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg

pipedream commented 13 years ago

comment:21

Replying to @dandrake:

As a data point: in a 32-bit virtual machine running the current version of Ubuntu 11.04 beta with 4.7.alpha4, the spkg here does build correctly and the crypt module imports as usual. All doctests pass, except that sage/sandpiles/sandpile.py timed out, and there's a minor rounding error in sage/matrix/matrix_double_dense.pyx -- neither seem related to the underlying Python spkg, so this looks good.

Do you mean the p9 spkg builds correctly and the fix is not necessary? Or the p10 builds correctly? Did you have the problem before with p9 or just test the p10 spkg after it was uploaded here? Does lsb_relaese -d give the output (replace first space with a tab):

Description: Ubuntu 11.04

jdemeyer commented 13 years ago

Merged: sage-4.7.rc1

jdemeyer commented 13 years ago

Changed merged from sage-4.7.rc1 to none

jdemeyer commented 13 years ago

comment:24

This breaks on Intel OS X 10.6 (bsd):

python-2.6.4.p10
Machine:
Darwin bsd.local 10.7.0 Darwin Kernel Version 10.7.0: Sat Jan 29 15:17:16 PST 2011; root:xnu-1504.9.37~1/RELEASE_I386 i386
Deleting directories from past builds of previous/current versions of python-2.6.4.p10
Extracting package /Users/buildbot/build/sage/bsd-1/bsd_full/build/sage-4.7.rc1/spkg/standard/python-2.6.4.p10.spkg ...
-rw-r--r--  1 buildbot  staff  11731636 Apr 24 05:34 /Users/buildbot/build/sage/bsd-1/bsd_full/build/sage-4.7.rc1/spkg/standard/python-2.6.4.p10.spkg
Finished extraction
****************************************************
Host system
uname -a:
Darwin bsd.local 10.7.0 Darwin Kernel Version 10.7.0: Sat Jan 29 15:17:16 PST 2011; root:xnu-1504.9.37~1/RELEASE_I386 i386
****************************************************
****************************************************
CC Version
gcc -v
Using built-in specs.
Target: i686-apple-darwin10
Configured with: /var/tmp/gcc/gcc-5666.3~6/src/configure --disable-checking --enable-werror --prefix=/usr --mandir=/share/man --enable-languages=c,objc,c++,obj-c++ --program-transform-name=/^[cg][^.-]*$/s/$/-4.2/ --with-slibdir=/usr/lib --build=i686-apple-darwin10 --program-prefix=i686-apple-darwin10- --host=x86_64-apple-darwin10 --target=i686-apple-darwin10 --with-gxx-include-dir=/include/c++/4.2.1
Thread model: posix
gcc version 4.2.1 (Apple Inc. build 5666) (dot 3)
****************************************************
[...]
gcc -dynamiclib -Wl,-single_module -L/Users/buildbot/build/sage/bsd-1/bsd_full/build/sage-4.7.rc1/local/lib  -undefined dynamic_lookup -Wl,-install_name,/Users/buildbot/build/sage/bsd-1/bsd_full/build/sage-4.7.rc1/local/lib/libpython2.6.dylib -Wl,-compatibility_version,2.6 -Wl,-current_version,2.6 -o libpython2.6.dylib Modules/getbuildinfo.o Parser/acceler.o Parser/grammar1.o Parser/listnode.o Parser/node.o Parser/parser.o Parser/parsetok.o Parser/bitset.o Parser/metagrammar.o Parser/firstsets.o Parser/grammar.o Parser/pgen.o Parser/myreadline.o Parser/tokenizer.o Objects/abstract.o Objects/boolobject.o Objects/bufferobject.o Objects/bytes_methods.o Objects/bytearrayobject.o Objects/cellobject.o Objects/classobject.o Objects/cobject.o Objects/codeobject.o Objects/complexobject.o Objects/descrobject.o Objects/enumobject.o Objects/exceptions.o Objects/genobject.o Objects/fileobject.o Objects/floatobject.o Objects/frameobject.o Objects/funcobject.o Objects/intobject.o Objects/iterobject.o Objects/listobject.o Objects/longobject.o Objects/dictobject.o Objects/methodobject.o Objects/moduleobject.o Objects/object.o Objects/obmalloc.o Objects/rangeobject.o Objects/setobject.o Objects/sliceobject.o Objects/stringobject.o Objects/structseq.o Objects/tupleobject.o Objects/typeobject.o Objects/weakrefobject.o Objects/unicodeobject.o Objects/unicodectype.o Python/_warnings.o Python/Python-ast.o Python/asdl.o Python/ast.o Python/bltinmodule.o Python/ceval.o Python/compile.o Python/codecs.o Python/errors.o Python/frozen.o Python/frozenmain.o Python/future.o Python/getargs.o Python/getcompiler.o Python/getcopyright.o Python/getmtime.o Python/getplatform.o Python/getversion.o Python/graminit.o Python/import.o Python/importdl.o Python/marshal.o Python/modsupport.o Python/mystrtoul.o Python/mysnprintf.o Python/peephole.o Python/pyarena.o Python/pyfpe.o Python/pymath.o Python/pystate.o Python/pythonrun.o Python/structmember.o Python/symtable.o Python/sysmodule.o Python/traceback.o Python/getopt.o Python/pystrcmp.o Python/pystrtod.o Python/formatter_unicode.o Python/formatter_string.o Python/dynload_shlib.o   Python/thread.o Modules/config.o Modules/getpath.o Modules/main.o Modules/gcmodule.o  Modules/threadmodule.o  Modules/signalmodule.o  Modules/posixmodule.o  Modules/errnomodule.o  Modules/pwdmodule.o  Modules/_sre.o  Modules/_codecsmodule.o  Modules/zipimport.o  Modules/symtablemodule.o  Modules/cryptmodule.o  Modules/xxsubtype.o -ldl    ; \

ar cr libpython2.6.a Modules/getbuildinfo.o
ar cr libpython2.6.a Parser/acceler.o Parser/grammar1.o Parser/listnode.o Parser/node.o Parser/parser.o Parser/parsetok.o Parser/bitset.o Parser/metagrammar.o Parser/firstsets.o Parser/grammar.o Parser/pgen.o Parser/myreadline.o Parser/tokenizer.o
ar cr libpython2.6.a Objects/abstract.o Objects/boolobject.o Objects/bufferobject.o Objects/bytes_methods.o Objects/bytearrayobject.o Objects/cellobject.o Objects/classobject.o Objects/cobject.o Objects/codeobject.o Objects/complexobject.o Objects/descrobject.o Objects/enumobject.o Objects/exceptions.o Objects/genobject.o Objects/fileobject.o Objects/floatobject.o Objects/frameobject.o Objects/funcobject.o Objects/intobject.o Objects/iterobject.o Objects/listobject.o Objects/longobject.o Objects/dictobject.o Objects/methodobject.o Objects/moduleobject.o Objects/object.o Objects/obmalloc.o Objects/rangeobject.o Objects/setobject.o Objects/sliceobject.o Objects/stringobject.o Objects/structseq.o Objects/tupleobject.o Objects/typeobject.o Objects/weakrefobject.o Objects/unicodeobject.o Objects/unicodectype.o
ar cr libpython2.6.a Python/_warnings.o Python/Python-ast.o Python/asdl.o Python/ast.o Python/bltinmodule.o Python/ceval.o Python/compile.o Python/codecs.o Python/errors.o Python/frozen.o Python/frozenmain.o Python/future.o Python/getargs.o Python/getcompiler.o Python/getcopyright.o Python/getmtime.o Python/getplatform.o Python/getversion.o Python/graminit.o Python/import.o Python/importdl.o Python/marshal.o Python/modsupport.o Python/mystrtoul.o Python/mysnprintf.o Python/peephole.o Python/pyarena.o Python/pyfpe.o Python/pymath.o Python/pystate.o Python/pythonrun.o Python/structmember.o Python/symtable.o Python/sysmodule.o Python/traceback.o Python/getopt.o Python/pystrcmp.o Python/pystrtod.o Python/formatter_unicode.o Python/formatter_string.o Python/dynload_shlib.o   Python/thread.o
/usr/bin/ranlib: file: libpython2.6.a(pymath.o) has no symbols
ar cr libpython2.6.a Modules/config.o Modules/getpath.o Modules/main.o Modules/gcmodule.o 
/usr/bin/ranlib: file: libpython2.6.a(pymath.o) has no symbols
ar cr libpython2.6.a Modules/threadmodule.o  Modules/signalmodule.o  Modules/posixmodule.o  Modules/errnomodule.o  Modules/pwdmodule.o  Modules/_sre.o  Modules/_codecsmodule.o  Modules/zipimport.o  Modules/symtablemodule.o  Modules/cryptmodule.o  Modules/xxsubtype.o
/usr/bin/ranlib: file: libpython2.6.a(pymath.o) has no symbols
ranlib libpython2.6.a
ranlib: file: libpython2.6.a(pymath.o) has no symbols
gcc -L/Users/buildbot/build/sage/bsd-1/bsd_full/build/sage-4.7.rc1/local/lib   -o python.exe \
            Modules/python.o \
            -L. -lpython2.6 -ldl  -lcrypt     
ld: library not found for -lcrypt
collect2: ld returned 1 exit status
make[2]: *** [python.exe] Error 1
Error building Python.

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:25

I'm puzzled by this, as the patch should only be applied on Ubuntu 10.04. Perhaps the "src" subdirectory got corrupted - I'll check it.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:26

Replying to @sagetrac-drkirkby:

I'm puzzled by this, as the patch should only be applied on Ubuntu 10.04. Perhaps the "src" subdirectory got corrupted - I'll check it.

Dave

That's the problem - the time stamp on the file 'Setup.dist' is recent, so it must have been changed.

-rw-r--r--   1 drkirkby other      28752 May 25  2010 _codecsmodule.c
-rw-r--r--   1 drkirkby other      21362 May 25  2010 _bytesio.c
-rw-r--r--   1 drkirkby other     207770 May 25  2010 _bsddb.c
-rw-r--r--   1 drkirkby other       6320 May 25  2010 _bisectmodule.c
-rw-r--r--   1 drkirkby other      18280 Apr 24 13:23 Setup.dist
drkirkby@laptop:~/python-2.6.4.p10/src/Modules$

This should be easy to fix. I'll make a new package, with the untouched 'Setup.dist'.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

Description changed:

--- 
+++ 
@@ -8,6 +8,6 @@

 Just merge 

-http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg
+http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:27

I've replaced the file src/Modules/Setup.dist which had got changed.

A new package can be found at

http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

Dave

jdemeyer commented 13 years ago

comment:28

Replying to @sagetrac-drkirkby:

I've replaced the file src/Modules/Setup.dist which had got changed.

A new package can be found at

http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg

Thanks! I will test it.

pipedream commented 13 years ago

comment:29

Sorry, that was me who tarred up the wrong Setup.dist. The new spkg from Dave does not build for me on Ubuntu 11.04. I am busy investigating why the patch did not apply.

pipedream commented 13 years ago

Description changed:

--- 
+++ 
@@ -6,8 +6,5 @@

 ## Note for the release manager

-Just merge 
+Not ready for merging

-http://boxen.math.washington.edu/home/kirkby/patches/python-2.6.4.p10.spkg
-
-

pipedream commented 13 years ago

comment:30

The sage subshell does not find lsb_release.

jan@mamadou-laptop:/usr/local/src/sage-4.6.2/spkg/build/python-2.6.4.p10$ cat mytest 
#!/usr/bin/env bash
lsb_release -d
jan@mamadou-laptop:/usr/local/src/sage-4.6.2/spkg/build/python-2.6.4.p10$ ./mytest 
Description:    Ubuntu 11.04
jan@mamadou-laptop:/usr/local/src/sage-4.6.2/spkg/build/python-2.6.4.p10$ sage -sh mytest 

Starting subshell with Sage environment variables set.
Be sure to exit when you are done and do not do anything
with other copies of Sage!

Bypassing shell configuration files ...

Traceback (most recent call last):
  File "/usr/bin/lsb_release", line 26, in <module>
    import lsb_release
ImportError: No module named lsb_release
Exited Sage subshell.
jan@mamadou-laptop:/usr/local/src/sage-4.6.2/spkg/build/python-2.6.4.p10$

pipedream commented 13 years ago

Changed reviewer from David Kirkby, Jan Groenewald to David Kirkby, Jan Groenewald, Dan Drake

pipedream commented 13 years ago

Description changed:

--- 
+++ 
@@ -1,8 +1,7 @@
 http://groups.google.com/group/sage-devel/browse_thread/thread/593b9a4124f5075d

-This patch is a fix for Ubuntu 11.04. Ubuntu already patch their python like this.  It is not a python bug. It is the Ubuntu+python combination, see the above thread for an answer I got from bugs.python.org.
-
-If this harms building on other systems, I can make it specific to checking the Ubuntu version and only then applying the patch (using lsb_release -d).
+This patch is a fix for Ubuntu 11.04. Ubuntu already patch their python like this.  It is not a python bug. It is the Ubuntu+python combination, see the above thread for an answer I got from bugs.python.org. It potentially affects future multiarch Debian/Ubuntu and derivatives. (At some point python 2.7 series
+may be patched upstream though.)

 ## Note for the release manager

pipedream commented 13 years ago

comment:31

New patch attached. lsb_release is a python script trying to import a system python package called lsb_release. I changed to check /etc/issue exists, and if so contains Ubuntu 11.04.

I also added an exit if the patch fails to apply, which we forgot to add before. That will also prevent already patched sources in src/ as the patch will fail to apply.

New spkg at http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg

pipedream commented 13 years ago

Attachment: 11243-crypt-module-Ubuntu11.04-issue.patch.gz

pipedream commented 13 years ago

Attachment: 11243-crypt-module-Ubuntu11.04-issue2.patch.gz

pipedream commented 13 years ago

comment:32

Attachment: 11243-crypt-module-Ubuntu11.04-issue3.patch.gz

All three "issue" patches need to be applied, in order. 1) change to /etc/issue instead of lsb_release 2) remove square brackets from grep test, 3) fix so it exits when patch fails, not always. This spkg does build correctly now on Ubuntu 11.04:

http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:33

I'm not sure why you have switched from using "lsb_release" to greping /etc/issue. It seems that "lsb_release" is a common command on most (all?) Linux systems, so the fact it works on your system's bash shell, but not in your Sage subshell, probably tells you the subshell or your test is broken. Why not fix that, rather than use another method?

As a general point, I would avoid the use of '-e' to test a file, as it's not portable. In this case it does not matter as the bash shell is invoked, but it would fail if you used a Bourne shell, so had:

#!/bin/sh

at the top.

From the man page on a Solaris system:

     -e file                     True if file exists. (Not avail-
                                 able in sh.)

Using '-f' is much safer as it will be found on far more shells.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:34

Oh, I see the problem. /usr/bin/lsb_release is a Python script, and in the Sage shell is picking up the Sage version of Python, not the system one. Got you. Now I see what that could be problematic, if Sage's Phython does not have all the bits for this script to work.

Dave

bac7d3ea-3f1b-4826-8464-f0b53d5e12d2 commented 13 years ago

comment:35

The error checking you added is not working correctly.

I tested this first on my OpenSolaris system - it worked OK.

Then I spoofed my machine to appear if it was Ubuntu 11.04. Your patch detected that, patched the python package, then it built OK. (Really whether it built or not was pretty irrelevant here, as I'd applied a Linux patch on Solaris).

Next as a test I corrupted the patch. When I try to install it, now I see:

Ubuntu 11.04
Patching src/Modules/Setup.dist as this is Ubuntu 11.04
patch: **** Only garbage was found in the patch input.
Error patching Setup.dist
Applying a revised 'configure' script for Solaris
See http://bugs.python.org/issue1759169
https://github.com/sagemath/sage-prod/issues/7867
Setting  HAVE_FD_TRANSFER=0 for Solaris to allow
the python module '_multiprocessing' to build
See: https://github.com/sagemath/sage-prod/issues/8440
checking for --with-universal-archs... 32-bit

In other words, despite the fact that Patch has reported the patch as garbage, the build just continues. So the error checking is not working correctly.

The man page for 'patch' says the exit code is 0 in the case of no error.

There is one other issue too, which is partially my fault. When I copied over the file you had corrupted, the time stamp on the directory has been changed. So it looks like the contents of {{src/Modules}}}' have been modified, even though they have not. It would be more sensible to copy the contents exactly, retaining all dates and times, so it does not leave anyone in any doubt the contents are unmodified.

I sometimes wonder if the files below the 'src' directory should be made read-only. That would stop accidentally changing them.

Dave

pipedream commented 13 years ago

comment:36

For compatibility with more shells, changed -e to -f. My previous error testing was in a subshell and I think exited that subshell only. I changed that to be like the exit of the other patches with an if statement checking the return value following the patch (or cp).

Patch attached and updated http://users.aims.ac.za/~jan/python-2.6.4.p10.spkg Builds on Ubuntu 11.04. Corrupted patch fails and exits correctly.

I'm not sure how to get the src/Modules/Setup.dist to the original timestamps. I tried untarring p9 and rsyncing -av Setup.dist but the date was not reset.

sagemath / sage

python spkg build fails on Ubuntu 11.04 #11243

Note for the release manager