Bug#695955: [Python-modules-team] Bug#695955: fail upon failures in loops + do not ignore tests failures

2013-01-03 Thread Sandro Tosi
Hello,
thanks for your work to make numpy a better package :)

On Fri, Dec 14, 2012 at 7:30 PM, Yaroslav Halchenko
deb...@onerussian.com wrote:
 Actually here are two patches (ready to be committed to SVN happen the
 maintainer approves)

 - one assures that all for loops are secured with set -e, so if any
 command fails, the loop fails (do not remember who pointed me to this drawback
 in my packages originally, thanks again ;) )

that makes sense, applied (in my local checkout, will commit once the
test build has finished).

 - tests failures are not ignored

they are * temporarily* ignored.

   imho it is much better to submit packages with only known to fail tests
   disabled/excluded or

 even better resolved

that's currently what's being done, cfr
https://github.com/numpy/numpy/issues/created_by/sandrotosi?page=1sort=updatedstate=open
+ closed.

 yes -- currently it would lead to FTBFS since all (or nearly all -- that
 is the point, impossible to figure out besides going through all of them 1 by
 1) architectures show some kind of test failure or a crash:

yes, I know, and as you probably know too, this patch won't be
accepted until all those tests are ironed out.

Regards,

--
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Bug#695955: [Python-modules-team] Bug#695955: Bug#695955: fail upon failures in loops + do not ignore tests failures

2013-01-03 Thread Yaroslav Halchenko

On Thu, 03 Jan 2013, Sandro Tosi wrote:
 thanks for your work to make numpy a better package :)

thanks for maintaining this core package ;)

  - one assures that all for loops are secured with set -e, so if any
  command fails, the loop fails (do not remember who pointed me to this 
  drawback
  in my packages originally, thanks again ;) )
 that makes sense, applied (in my local checkout, will commit once the
 test build has finished).

coolio

  - tests failures are not ignored
 they are * temporarily* ignored.

yeah -- later on Julian hinted me on that ;-)

imho it is much better to submit packages with only known to fail tests
disabled/excluded or
  even better resolved
 that's currently what's being done, cfr
 https://github.com/numpy/numpy/issues/created_by/sandrotosi?page=1sort=updatedstate=open
 + closed.

cool!

  yes -- currently it would lead to FTBFS since all (or nearly all -- that
  is the point, impossible to figure out besides going through all of them 1 
  by
  1) architectures show some kind of test failure or a crash:
 yes, I know, and as you probably know too, this patch won't be
 accepted until all those tests are ironed out.

fair enough

thanks again
-- 
Yaroslav O. Halchenko
Postdoctoral Fellow,   Department of Psychological and Brain Sciences
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834   Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Bug#695955: fail upon failures in loops + do not ignore tests failures

2012-12-14 Thread Yaroslav Halchenko
Package: python-numpy
Version: 1:1.6.2-1
Severity: wishlist
Tags: patch

Actually here are two patches (ready to be committed to SVN happen the
maintainer approves)

- one assures that all for loops are secured with set -e, so if any
command fails, the loop fails (do not remember who pointed me to this drawback
in my packages originally, thanks again ;) )

- tests failures are not ignored

  imho it is much better to submit packages with only known to fail tests
  disabled/excluded or even better resolved -- that provides some guarantee
  that there is no unknown issues left

yes -- currently it would lead to FTBFS since all (or nearly all -- that
is the point, impossible to figure out besides going through all of them 1 by
1) architectures show some kind of test failure or a crash:

https://buildd.debian.org/status/package.php?p=python-numpysuite=experimental

-- System Information:
Debian Release: wheezy/sid
  APT prefers testing
  APT policy: (900, 'testing'), (600, 'unstable'), (300, 'experimental'), (100, 
'stable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 3.2.0-3-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages python-numpy depends on:
ii  libatlas3-base [liblapack.so.3gf]  3.8.4-7
ii  libatlas3gf-base   3.8.4-7
ii  libblas3 [libblas3gf]  1.2.20110419-5
ii  libblas3gf 1.2.20110419-5
ii  libc6  2.13-33
ii  libgcc11:4.7.2-4
ii  libgfortran3   4.7.2-4
ii  liblapack3 [liblapack3gf]  3.4.1-4
ii  liblapack3gf   3.4.1-4
ii  libquadmath0   4.7.2-4
ii  python 2.7.3-3
ii  python-support 1.0.15

python-numpy recommends no packages.

Versions of packages python-numpy suggests:
ii  gcc   4:4.7.2-1
ii  gfortran  4:4.7.1-1
ii  python-dev2.7.3-3
ii  python-nose   1.1.2-3
pn  python-numpy-dbg  none
ii  python-numpy-doc  1:1.6.2-1

-- no debconf information
From 46bbfedb75f53f8f89ea749c60a4fc7c1ed75e3d Mon Sep 17 00:00:00 2001
From: Yaroslav Halchenko deb...@onerussian.com
Date: Fri, 14 Dec 2012 13:10:34 -0500
Subject: [PATCH 1/2] debian/rules:  safe-guard all for loops with 'set -e; '
 to prevent uncaught failures

---
 debian/changelog |  7 +++
 debian/rules | 32 
 2 files changed, 23 insertions(+), 16 deletions(-)

diff --git a/debian/changelog b/debian/changelog
index b701e8a..2f5ca09 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,3 +1,10 @@
+python-numpy (1:1.7.0~b2-2) UNRELEASED; urgency=low
+
+  * debian/rules
+- safe-guard all for loops with 'set -e; ' to prevent uncaught failures
+
+ -- Yaroslav Halchenko deb...@onerussian.com  Fri, 14 Dec 2012 13:08:56 -0500
+
 python-numpy (1:1.7.0~b2-1) experimental; urgency=low
 
   * New upstream beta release
diff --git a/debian/rules b/debian/rules
index ca48806..fc3d41c 100755
--- a/debian/rules
+++ b/debian/rules
@@ -16,7 +16,7 @@ export ATLAS=None
 
 override_dh_auto_build:
 	dh_auto_build
-	for v in $(PY3VERS); do \
+	set -e; for v in $(PY3VERS); do \
 		python$$v setup.py build; \
 		python$$v-dbg setup.py build; \
 	done
@@ -34,11 +34,11 @@ override_dh_installman:
 	mv debian/python3-numpy/usr/share/man/man1/f2py.1 \
 	  debian/python3-numpy/usr/share/man/man1/f2py3.1
 	# link manpage for versioned and dbg incarnations of f2py
-	for v in $(PY2VERS); do \
+	set -e; for v in $(PY2VERS); do \
 		dh_link /usr/share/man/man1/f2py.1.gz /usr/share/man/man1/f2py$$v.1.gz; \
 		dh_link -ppython-numpy-dbg /usr/share/man/man1/f2py.1.gz /usr/share/man/man1/f2py$$v-dbg.1.gz; \
 	done
-	for v in $(PY3VERS); do \
+	set -e; for v in $(PY3VERS); do \
 		dh_link -ppython3-numpy /usr/share/man/man1/f2py3.1.gz /usr/share/man/man1/f2py$$v.1.gz; \
 		dh_link -ppython3-numpy-dbg /usr/share/man/man1/f2py3.1.gz /usr/share/man/man1/f2py$$v-dbg.1.gz; \
 	done
@@ -47,7 +47,7 @@ override_dh_installman:
 
 override_dh_install:
 	# add shebang information to f2py script
-	for v in $(PY2VERS) $(PY3VERS); do \
+	set -e; for v in $(PY2VERS) $(PY3VERS); do \
 		sed -i 1s,#!.*python[^ ]*\(.*\),#!/usr/bin/python$$v, debian/tmp/usr/bin/f2py$$v; \
 		cp -a debian/tmp/usr/bin/f2py$$v debian/tmp/usr/bin/f2py$$v-dbg ; \
 		sed -i 1s,#!.*python[^ ]*\(.*\),#!/usr/bin/python$$v-dbg, debian/tmp/usr/bin/f2py$$v-dbg; \
@@ -74,23 +74,23 @@ override_dh_install:
 	find $(CURDIR)/debian/python-numpy/ -name *_d.so -delete
 
 	# create symlinks for .h files
-	for i in $(PY2VERS); do \
+	set -e; for i in $(PY2VERS); do \
 	[ -d $(CURDIR)/debian/python-numpy/usr/include/python$$i ] || \
 		mkdir -p $(CURDIR)/debian/python-numpy/usr/include/python$$i; \
 		dh_link usr/lib/pymodules/python$$i/numpy/core/include/numpy usr/include/python$$i/numpy; \
 	done
-	for i in $(PY2VERS); do \
+	set -e; for