Re: Official position on POSIX compliance?
At Tue, 6 Jul 2004 22:46:33 -0400 (EDT),
Jurij Smakov wrote:
> I am not a developer. However, currently I am preparing a package and
> do some random bug fixing. At some point that brought me across the
> issue of POSIX compliance. Even though I've got the impression (from
> mailing lists and IRC) that this is The Right Thing (tm), I was not able
> to find either any official statement, or a statement expressing the
> consensus of Debian developers on the issue. Maybe, such document(s)
> exist, in which case I would appreciate the pointers. Assuming, that it
> does not, I would like to ask a few questions, to get the discussion
> going, and, possibly, produce such a document.
> For those not very familiar with the issue, POSIX is Portable Operating
> System Interface, a standard developed jointly by the IEEE and The Open
> Group. Between other things, it defines the shell syntax and options,
> which should be supported by POSIX-compliant utilities . There is quite
> a lot of maintainer scripts out there, which require minor modifications
> to make them POSIX-compliant. Most frequently encountered cases are:
There are various POSIX related standards. I guess your term "POSIX
compliant" is only for XBD and XCU in 1003.1-200x.
> * user/group specification as 'user.group' as an argument for chown, for
> example. The correct POSIX form is 'user:group'.
> * Arguments -a and -o to 'test', which are XSI extensions.
IIRC, there were some bugs to annoy this kind of issue. Search BTS.
> The questions which I would like to have answered are:
> * Does the majority of Debian developers agree, that this compliance
> should be enforced and bugs should be filed against the non-compliant
Yes it should be.
> * Why isn't POSIX compliance commitment mentioned anywhere in the policy
> (except section 10.4, which mentions that all scripts, executed under
> /bin/sh must be POSIX-compliant)?
It's good idea to follow the POSIX standards. But satisfying POSIX
compliance completely is difficult. Note that LSB includes the
pointer to POSIX standards.
> * What severity should be assigned to non-POSIX-compliance bugs (if any)?
I think it's changed bugs' importance; from serious to wishlist.
Sometimes satisfying POSIX needs a large amount of updates including
kernel modifications (one recent example is complete pthread support).
It should be wishlist. OTOH, "chmod user.group" is serious; it may
break on some environments.