Report bugs for variations in error handling in math functions?

Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

Currently, I'm revising all of the math pages in man-pages, and in the
process testing the error handling (glibc 2.8) for each function.

I find the following:

a) on error, many (probably a majority of) functions set errno AND
raise an exception (fetestexcept()).
b) on error, a very few functions DO set errno but DON"T raise an
exception (fetestexcept()).
c) on error, a few functions DON'T set errno but DO raise an exception
(fetestexcept()).
d) on error, a very few functions pursue a mixture of all of the
above, depending on the error.

A math_error(7) page that I recently wrote (see
http://www.kernel.org/doc/man-pages/online/pages/man7/math_error.7.html
) currently implies that all functions should do a).  Clearly I'll
need to amend that.

But the main question is, should I raise glibc bugs for the functions
in cases b), c), and d)?

Cheers,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Joseph S. Myers]

On Fri, 25 Jul 2008, Michael Kerrisk wrote:

> Currently, I'm revising all of the math pages in man-pages, and in the
> process testing the error handling (glibc 2.8) for each function.
> 
> I find the following:
> 
> a) on error, many (probably a majority of) functions set errno AND
> raise an exception (fetestexcept()).
> b) on error, a very few functions DO set errno but DON"T raise an
> exception (fetestexcept()).
> c) on error, a few functions DON'T set errno but DO raise an exception
> (fetestexcept()).
> d) on error, a very few functions pursue a mixture of all of the
> above, depending on the error.
> 
> A math_error(7) page that I recently wrote (see
> http://www.kernel.org/doc/man-pages/online/pages/man7/math_error.7.html
> ) currently implies that all functions should do a).  Clearly I'll
> need to amend that.
> 
> But the main question is, should I raise glibc bugs for the functions
> in cases b), c), and d)?

I've run third-party C conformance tests on glibc that have shown similar 
issues.

C90 requires errno to be set by various functions.  C99 allows it not to 
be set, in an incompatible quiet change from C90, if exceptions are used 
instead.  The correct handling under C99 depends on the value of 
math_errhandling.  Implementing math_errhandling requires compiler and 
linker support (see messages linked from the CONFORMANCE file): if any 
translation unit is compiled with -fno-math-errno then math_errhandling & 
MATH_ERRNO must not be set.  (I'd suggest that the compiler set object 
attributes which the static linker then uses to provide the relevant 
information to libc.)

I think the correct approach is to consider it a bug if functions do not 
set errno, or do not raise exceptions - that is, all of (b), (c) and (d) 
are bugs.  This would allow math_errhandling to be 
MATH_ERRNO|MATH_ERREXCEPT unless translation units are compiled with 
options preventing this, and make the error handling options available 
consistent across the math.h functions supported by glibc.  To conform 
with C99, at least one approach (errno or exceptions) must be consistently 
supported across all the functions, in any case.

-- 
Joseph S. Myers
joseph-qD8j1LwMmJjtCj0u4l0SBw@public.gmane.org
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

On Fri, Jul 25, 2008 at 5:43 PM, Joseph S. Myers
<joseph-qD8j1LwMmJjtCj0u4l0SBw@public.gmane.org> wrote:
> On Fri, 25 Jul 2008, Michael Kerrisk wrote:
>
>> Currently, I'm revising all of the math pages in man-pages, and in the
>> process testing the error handling (glibc 2.8) for each function.
>>
>> I find the following:
>>
>> a) on error, many (probably a majority of) functions set errno AND
>> raise an exception (fetestexcept()).
>> b) on error, a very few functions DO set errno but DON"T raise an
>> exception (fetestexcept()).
>> c) on error, a few functions DON'T set errno but DO raise an exception
>> (fetestexcept()).
>> d) on error, a very few functions pursue a mixture of all of the
>> above, depending on the error.
>>
>> A math_error(7) page that I recently wrote (see
>> http://www.kernel.org/doc/man-pages/online/pages/man7/math_error.7.html
>> ) currently implies that all functions should do a).  Clearly I'll
>> need to amend that.
>>
>> But the main question is, should I raise glibc bugs for the functions
>> in cases b), c), and d)?
>
> I've run third-party C conformance tests on glibc that have shown similar
> issues.
>
> C90 requires errno to be set by various functions.  C99 allows it not to
> be set, in an incompatible quiet change from C90, if exceptions are used
> instead.  The correct handling under C99 depends on the value of
> math_errhandling.  Implementing math_errhandling requires compiler and
> linker support (see messages linked from the CONFORMANCE file):

Yes, I saw those already.

> if any
> translation unit is compiled with -fno-math-errno then math_errhandling &
> MATH_ERRNO must not be set.  (I'd suggest that the compiler set object
> attributes which the static linker then uses to provide the relevant
> information to libc.)
>
> I think the correct approach is to consider it a bug if functions do not
> set errno, or do not raise exceptions - that is, all of (b), (c) and (d)
> are bugs.  This would allow math_errhandling to be
> MATH_ERRNO|MATH_ERREXCEPT unless translation units are compiled with
> options preventing this, and make the error handling options available
> consistent across the math.h functions supported by glibc.  To conform
> with C99, at least one approach (errno or exceptions) must be consistently
> supported across all the functions, in any case.

Not even one approach is consistent.  Many functions don't set errno.
But there are one or two that don't raise exceptions: the Bessel
functions (j0(), y0 etc), for example.  It's all a bit of a mess,
unfortunately.

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Joseph S. Myers]

On Fri, 25 Jul 2008, Michael Kerrisk wrote:

> Not even one approach is consistent.  Many functions don't set errno.
> But there are one or two that don't raise exceptions: the Bessel
> functions (j0(), y0 etc), for example.  It's all a bit of a mess,
> unfortunately.

Those functions aren't in C99.  They are however in POSIX, which defines 
C99-like error handling requirements, though a lot of errors are "may 
occur" just as a lot of C99 ones are.  (But if a "may occur" error is 
indicated by one of errno and exceptions, it must be indicated by both if 
math_errhandling says both forms of error handling are supported.)

(Bessel functions under different names and a lot of other functions are 
in DIS 24747, which passed its ballot for approval as an International 
Standard.  But I don't expect it to be integrated in the next revision of 
the C standard; my understanding is also that all of C++ TR1 *apart* from 
special mathematical functions (much the same set) is being integrated in 
C++0x.)

-- 
Joseph S. Myers
joseph-qD8j1LwMmJjtCj0u4l0SBw@public.gmane.org
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

On Fri, Jul 25, 2008 at 5:43 PM, Joseph S. Myers
<joseph-qD8j1LwMmJjtCj0u4l0SBw@public.gmane.org> wrote:
> On Fri, 25 Jul 2008, Michael Kerrisk wrote:
>
>> Currently, I'm revising all of the math pages in man-pages, and in the
>> process testing the error handling (glibc 2.8) for each function.
>>
>> I find the following:
>>
>> a) on error, many (probably a majority of) functions set errno AND
>> raise an exception (fetestexcept()).
>> b) on error, a very few functions DO set errno but DON"T raise an
>> exception (fetestexcept()).
>> c) on error, a few functions DON'T set errno but DO raise an exception
>> (fetestexcept()).
>> d) on error, a very few functions pursue a mixture of all of the
>> above, depending on the error.
>>
>> A math_error(7) page that I recently wrote (see
>> http://www.kernel.org/doc/man-pages/online/pages/man7/math_error.7.html
>> ) currently implies that all functions should do a).  Clearly I'll
>> need to amend that.
>>
>> But the main question is, should I raise glibc bugs for the functions
>> in cases b), c), and d)?
>
> I've run third-party C conformance tests on glibc that have shown similar
> issues.

By the way, Joseph, which particular conformance test suite(s) were
you using?  Are these freely available test suite(s)?
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Joseph S. Myers]

On Mon, 28 Jul 2008, Michael Kerrisk wrote:

> > I've run third-party C conformance tests on glibc that have shown similar
> > issues.
> 
> By the way, Joseph, which particular conformance test suite(s) were
> you using?  Are these freely available test suite(s)?

No, proprietary (Plum Hall).  I'm not aware of free C library conformance 
testsuites, although such tests may be available in particular areas (e.g. 
I used ucbtest to test soft-fp changes).

-- 
Joseph S. Myers
joseph-qD8j1LwMmJjtCj0u4l0SBw@public.gmane.org
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

> I think the correct approach is to consider it a bug if functions do not 
> set errno, or do not raise exceptions - that is, all of (b), (c) and (d) 
> are bugs.  This would allow math_errhandling to be 
> MATH_ERRNO|MATH_ERREXCEPT unless translation units are compiled with 
> options preventing this, and make the error handling options available 
> consistent across the math.h functions supported by glibc.  To conform 
> with C99, at least one approach (errno or exceptions) must be consistently 
> supported across all the functions, in any case.

I've reported bugs for all of the non-conformances I found.

Cheers,

Michael

http://sources.redhat.com/bugzilla/show_bug.cgi?id=6759
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6776
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6777
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6778
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6779
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6780
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6781
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6782
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6783
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6783
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6783
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6785
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6786
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6787
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6788
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6792
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6793
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6794
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6795
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6796
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6797
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6798
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6799
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6801
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6801
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6803
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6803
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6804
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6805
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6806
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6807
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6808
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6809
http://sources.redhat.com/bugzilla/show_bug.cgi?id=6810

--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

On Thu, Jul 31, 2008 at 4:02 PM, Michael Kerrisk
<mtk.manpages-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org> wrote:
>> I think the correct approach is to consider it a bug if functions do not
>> set errno, or do not raise exceptions - that is, all of (b), (c) and (d) are
>> bugs.  This would allow math_errhandling to be MATH_ERRNO|MATH_ERREXCEPT
>> unless translation units are compiled with options preventing this, and make
>> the error handling options available consistent across the math.h functions
>> supported by glibc.  To conform with C99, at least one approach (errno or
>> exceptions) must be consistently supported across all the functions, in any
>> case.
>
> I've reported bugs for all of the non-conformances I found.

What I really should have said here is "non-conformances and
deviations from the/my ideal that all errors would be reported both
via errno and exceptions".  Anyway, one for the list:

http://sources.redhat.com/bugzilla/show_bug.cgi?id=6814
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Report bugs for variations in error handling in math functions?

[Michael Kerrisk]

On Mon, Aug 4, 2008 at 7:48 AM, Michael Kerrisk
<mtk.manpages-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org> wrote:
> On Thu, Jul 31, 2008 at 4:02 PM, Michael Kerrisk
> <mtk.manpages-gM/Ye1E23mwN+BqQ9rBEUg@public.gmane.org> wrote:
>>> I think the correct approach is to consider it a bug if functions do not
>>> set errno, or do not raise exceptions - that is, all of (b), (c) and (d) are
>>> bugs.  This would allow math_errhandling to be MATH_ERRNO|MATH_ERREXCEPT
>>> unless translation units are compiled with options preventing this, and make
>>> the error handling options available consistent across the math.h functions
>>> supported by glibc.  To conform with C99, at least one approach (errno or
>>> exceptions) must be consistently supported across all the functions, in any
>>> case.
>>
>> I've reported bugs for all of the non-conformances I found.
>
> What I really should have said here is "non-conformances and
> deviations from the/my ideal that all errors would be reported both
> via errno and exceptions".  Anyway, one for the list:
>
> http://sources.redhat.com/bugzilla/show_bug.cgi?id=6814

A tabular summary of what I found is provided here:

http://linux-man-pages.blogspot.com/2008/08/math-functions-and-error-reporting.html

Cheers,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html