Identify modules which couldn't be built

[smontanaro]

BPO	1673619
Nosy	@loewis, @smontanaro
Files	missing.diff

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/smontanaro'
closed_at = <Date 2007-03-06.15:42:45.000>
created_at = <Date 2007-03-04.21:31:41.000>
labels = ['library']
title = "Identify modules which couldn't be built"
updated_at = <Date 2007-03-06.15:42:45.000>
user = 'https://github.com/smontanaro'

bugs.python.org fields:

activity = <Date 2007-03-06.15:42:45.000>
actor = 'skip.montanaro'
assignee = 'skip.montanaro'
closed = True
closed_date = None
closer = None
components = ['Distutils']
creation = <Date 2007-03-04.21:31:41.000>
creator = 'skip.montanaro'
dependencies = []
files = ['7822']
hgrepos = []
issue_num = 1673619
keywords = ['patch']
message_count = 6.0
messages = ['52046', '52047', '52048', '52049', '52050', '52051']
nosy_count = 2.0
nosy_names = ['loewis', 'skip.montanaro']
pr_nums = []
priority = 'normal'
resolution = 'accepted'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue1673619'
versions = ['Python 2.6']

[smontanaro]

A user has been moaning on c.l.py that Python's setup.py doesn't bail out when the necessary bits can't be found to build a particular extension module (curses in this case). For reasons that should be obvious to this crowd you simply can't have it stop if a particular module can't be built.

This patch is a compromise. It enumerates the modules for which the necessary build mojo exists and those modules for which it doesn't. I'm personally a bit ambivalent about it, both the need for it and my particular implementation. An alternate implementation might create a set of all module names at the start of the run then subtract the modules which could be built from that. This implementation has the slight advantage that all the fiddling for a particular module occurs near the tests for that module.

Skip

[loewis]

I don't think the modules that it builds successfully need to be printed - the build messages will tell, anyway.

I'm not sure when it prints these things. IMO, they should be printed at the end of the build (i.e. after all modules have been built). It should also print out, in summary, modules that had been attempted to build, but where building failed for some reason.

[smontanaro]

Martin,

Thanks for the feedback. Attached is an updated diff. It defers the
report until build_ext is finished and reports only on what couldn't
be built, either because the environment was missing something or
because the compile/link phase failed. These two different problems
are reported in different paragraphs, e.g.:

Failed to find the necessary bits to build these modules:
_md5              _sha              _sha256
_sha512           gdbm              linuxaudiodev
ossaudiodev       spwd              sunaudiodev


Failed to build these modules:
_elementtree

File Added: missing.diff

[smontanaro]

Tweak summary.

[loewis]

Some nits:

• the max computation can be simplified as
  longest = max(len(e.name) for e in self.extensions+self.failed)

• The sort can be simplified as
  lst.sort(key=str.lower)

There is a change to xmlrpclib in this patch that probably isn't meant to be part of it.

Otherwise, it looks fine, please apply.

[smontanaro]

Checked in as r54166. I applied the sort() change but not the
max() change. Note that self.extensions is a list of Extension instances while self.failed is a list of strings.