[Bug web/35777] New: no DFP announcement, no example text, very vague documentation
bkoz at gcc dot gnu dot org
gcc-bugzilla@gcc.gnu.org
Mon Mar 31 18:40:00 GMT 2008
Filing this so that the multiple requests to add DFP docs don't continue to be
ignored.
First asked here:
http://gcc.gnu.org/ml/gcc/2007-06/msg00048.html
Asked again 9 months later:
http://gcc.gnu.org/ml/gcc/2008-03/msg01025.html
Janis indicates that some patch was proposed, but then not applied for lack of
examples. I suggest instead a gradual approach, where something is applied
immediately to gcc-4.3/changes.html and then refined.
See:
http://gcc.gnu.org/ml/gcc-patches/2007-09/msg01139.html
For a starting point.
My preference would be for the examples and more extensive usage docs to go on
to the main manual node for DFP, and the announcement to link to that page.
The lapse in documenting this feature, after repeated requests to do so, is
highly annoying. GCC is a large and fast-moving target: it is hard for users
and maintainers to understand the whole picture without the good-faith efforts
of developers to do due diligence and document new work. Significant new
features or experimental modes should be documented to give all maintainers an
equal foothold, and users a fighting chance to keep up with compiler
technology.
As it stands, DFP is not documented in changes.html, but *changes* to
*existing* DFP support *IS* mentioned. This is inconsistent.
In addition to the announcement text, the GCC manual is quite light on DFP
details:
Main DFP doc node:
http://gcc.gnu.org/onlinedocs/gcc/Decimal-Float.html#Decimal-Float
Does this need to be updated?
If DFP support and standard are a moving target, then it may be beneficial to
just add a dfp_status.html page, and update that in a similar manner to other
experimental support (like C++0x).
And the other hit:
http://gcc.gnu.org/onlinedocs/gcc/Option-Summary.html#Option-Summary
lists:
-mno-mfpgpr -mhard-dfp
for RS/6000 only. Documentation of this is:
The -mhard-dfp option allows GCC to generate the decimal floating point
instructions implemented on some POWER processors.
This is pretty vague, and I believe incorrect if DFP is now supported on x86,
as recent gcc-patches activity implies.
Timely action on behalf of DFP maintainers will be appreciated.
best,benjamin
--
Summary: no DFP announcement, no example text, very vague
documentation
Product: gcc
Version: 4.3.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: web
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: bkoz at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35777
More information about the Gcc-bugs
mailing list