make html in doc fails because Makefile assigns python to PYTHON

[MLModel]

BPO	5658
Nosy	@birkenfeld, @MLModel
Files	unnamed unnamed unnamed

^{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 = None
closed_at = <Date 2009-04-01.23:11:19.541>
created_at = <Date 2009-04-01.21:09:52.894>
labels = ['build']
title = 'make html in doc fails because Makefile assigns python to \tPYTHON'
updated_at = <Date 2009-04-02.17:36:45.334>
user = 'https://github.com/MLModel'

bugs.python.org fields:

activity = <Date 2009-04-02.17:36:45.334>
actor = 'MLModel'
assignee = 'none'
closed = True
closed_date = <Date 2009-04-01.23:11:19.541>
closer = 'georg.brandl'
components = ['Build']
creation = <Date 2009-04-01.21:09:52.894>
creator = 'MLModel'
dependencies = []
files = ['13565', '13581', '13583']
hgrepos = []
issue_num = 5658
keywords = []
message_count = 7.0
messages = ['85094', '85113', '85122', '85147', '85204', '85208', '85227']
nosy_count = 2.0
nosy_names = ['georg.brandl', 'MLModel']
pr_nums = []
priority = 'normal'
resolution = 'wont fix'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue5658'
versions = ['Python 3.1']

[MLModel]

Mac OSX, py3k 70991
cd doc; make html
fails due to an old style except clause in Doc/tools/sphinx/init.py

The make works in release30-maint.

The difference is that the generated doc/Makefile in release30-maint
assigns python2.5 to PYTHON, but the py3k doc/Makefile assigns python. I
have things set up so that 'python' is Python 3.

I understand how to set the variable from the command line, and I can
build the html doc, but this should be fixed, either by continuing to
specify python2.5 in the sphinx Makefile or by changing __init__.py and
whatever other code uses exception binding to the new syntax.

[birkenfeld]

We don't recommend calling Python 3.x "python". In fact, the executable
will always be called "python3" in Python 3.1+.

[MLModel]

OK, makes sense. Just out of curiosity where does a user find that
information? a developer?

There is still an inconsistency between the Makefiles in the two
versions. That seems dangerous.

--

     --- Mitchell

[birkenfeld]

Why is a change to the Makefile an "inconsistency"? Anyway, once 3.1 is
released the naming scheme will be announced.

[MLModel]

It strikes me as an inconsistency because I wouldn't expect the build
process to differ from Python 3.0 to 3.1 except for fixing bugs. A
developer, or someone trying to keep up with the most recent
versions, who had been successfiully making the html doc in a
subversion copy of Python 3.0 and wanted to move over to Python 3.1
finds this surprise if the command python is bound to Python 3. I
grant you that this is a lot of ifs, and a transitional issue if
we're all going to call Python 3 python3.

It might even be worth mentioning in the build instructions that
"python" should be bound to a Python 2 executable, not Python 3,
because a few things, such as make html in doc, use tools that
haven't been updated.

Not a big deal. Just fighting for the "Principle of Least Surprise",
as in many of the documentation issues I've submitted. My GUI
background showing.

--

     --- Mitchell

[birkenfeld]

Even Python 3.0 didn't install its built executable as "python". We
just made that decision, which was done for Python 3.0 because it
couldn't be considered stable, permanent.

[MLModel]

OK

--

     --- Mitchell