This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

classification
Title: --without-cxx flag of configure isn't documented.
Type: Stage:
Components: Documentation Versions: Python 2.2
process
Status: closed Resolution: out of date
Dependencies: Superseder:
Assigned To: Nosy List: brett.cannon, facundobatista, jschonek2, loewis
Priority: normal Keywords:

Created on 2003-03-12 12:05 by jschonek2, last changed 2022-04-10 16:07 by admin. This issue is now closed.

Messages (9)
msg15090 - (view) Author: J.A. Schonekerl (jschonek2) Date: 2003-03-12 12:05
Hi,

I can't any discription of the --without-cxx for configure.
If you do a ./configure --help it simply doesn't show up!

And let this be the flag I needed to build python without
libstdc++!

Best regards,

Jan
msg15091 - (view) Author: Martin v. Löwis (loewis) * (Python committer) Date: 2003-03-15 13:33
Logged In: YES 
user_id=21627

This is a general configure principle: For every --with
option, it is possible to pass --without, and for every
--enable option, it is posible to pass --disable.

Would you be willing to create a patch that explains this to
the user in a way that you would understand?
msg15092 - (view) Author: J.A. Schonekerl (jschonek2) Date: 2003-03-17 11:49
Logged In: YES 
user_id=732192


For me it wasn't clear why the libstdc++ library was dynamicaly 
linked into the python interpreter. I could not find the
relation 
between the --with-cxx=<compiler> flag and the libstdc++
library. The README file tells something about the --with-cxx
flag. But not that this causes the libstdc++ inclusion (not
everyone is C/C++ expert). Afther reading the README file it's
also not logical why --with-cxx is choosen as default.

Another problem a had:
I'am missing the default settings with the --with/--wthout and 
--disable/--enable settings, or are all settings listed under 
"--enable and --with options recognized:" in the configure
output defaults? 
Maybe this info should be added the same way as this line
from the configure output.
"Options: [defaults in brackets after descriptions]".

I have these points for improvement:

- Make clear in the README file, with the --with-cxx
explenation that this causes libstdc++ inclusion.

- Make clear in the configure output that all listed
--with/--without/--enable/--disable flags are the default (mmm,
this sounds logical for me now...).

- Make clear in the configure output that all listed
--with/--without/--enable/--disable flags have always a
oposite. (this sounds also logical now, they're all bools....).

- Why is --with-cxx choosen as default, the README file
suggests that you only want this if..............

I'am not a ./configure or c++ expert, maybe my points are
logical
and normal behaviour for you. But I'am sure they will help
other poeple that arn't expert either.

And yes I can make a patch for these points.
msg15093 - (view) Author: J.A. Schonekerl (jschonek2) Date: 2003-03-17 12:24
Logged In: YES 
user_id=732192

For example..........

The postgresql configure help does it better:

Optional Packages:
  --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
  --without-PACKAGE       do not use PACKAGE (same as
--with-PACKAGE=no)

Optional Features:
  --disable-FEATURE       do not include FEATURE (same as
--enable-FEATURE=no)
  --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]

It also adds:

Some influential environment variables:
  CC          C compiler command
  CFLAGS      C compiler flags
  LDFLAGS     linker flags, e.g. -L<lib dir> if you have
libraries in a
              nonstandard directory <lib dir>
  CPPFLAGS    C/C++ preprocessor flags, e.g. -I<include dir>
if you have
              headers in a nonstandard directory <include dir>
  CPP         C preprocessor
  DOCBOOKSTYLE
              location of DocBook stylesheets

But if all listed "oiptional pakages" and "optional
features" are
the default isn't clear in this configure output either.

The Python configure needs a general face-lift.
msg15094 - (view) Author: Martin v. Löwis (loewis) * (Python committer) Date: 2003-03-17 18:39
Logged In: YES 
user_id=21627

Being a C++ expert won't help you in explaining libstdc++
dependencies: This is specific to GNU C++, not to C++ in
general. So you need to be an expert of your operating
system; to some degree, this is expected from somebody
building software on a system.

I *think* that configure --help lists not the defaults, but
the non-defaults in most cases, i.e. the options that you
need to pass. In some cases (e.g. --with-cxx=), they are not
really booleans, but, if activated, also support a parameter
(the name of the C++ compiler, for --with-cxx). In that
case, the default is more involved: If not specified,
--with-cxx is assumed if a C++ compiler can be found and if
linking with that C++ compiler is necessary on your system
(some systems support C++ extension but don't require Python
to be linked with the C++ compiler).

So, please do propose specific wording. I think there is
little point in explaining every detail of what exactly the
configuration tries in what circumstance; so maybe something
general enough that still would have helped you might be
appropriate.
msg15095 - (view) Author: J.A. Schonekerl (jschonek2) Date: 2003-03-17 19:30
Logged In: YES 
user_id=732192

It becomes clear for me that making a good help for all 
systems out there is very difficult. 

But I could create a better ./configure --help output for python
that is comon for all systems.

And I can create a patch for the README file that explains
the libstdc++ dependencies if you are building with GNU C++.

Can I simply add a patch for these to the patch list or how 
does it work?
msg15096 - (view) Author: Martin v. Löwis (loewis) * (Python committer) Date: 2003-03-17 22:14
Logged In: YES 
user_id=21627

Yes, please attached unified or context diffs to this bug
report, preferably using the CVS version of Python (or 2.3a2
if CVS is not possible).
msg15097 - (view) Author: Facundo Batista (facundobatista) * (Python committer) Date: 2005-01-15 12:53
Logged In: YES 
user_id=752496

Please, could you verify if this problem persists in Python 2.3.4
or 2.4?

If yes, in which version? Can you provide a test case?

If the problem is solved, from which version?

Note that if you fail to answer in one month, I'll close this bug
as "Won't fix".

Thank you! 

.    Facundo
msg15098 - (view) Author: Brett Cannon (brett.cannon) * (Python committer) Date: 2005-01-17 00:26
Logged In: YES 
user_id=357491

This is no longer a problem.  Closed as "out of date".
History
Date User Action Args
2022-04-10 16:07:35adminsetgithub: 38151
2003-03-12 12:05:47jschonek2create