[lsb-discuss] "Standardizing" 7 LSB HW architecture names
vrub at ispras.ru
Fri Mar 28 06:42:36 PDT 2008
when working on LSB in many places we need to indicate names of HW
architectures (the 7 ones LSB currently supports). And there is
inconsistency in exact naming of the architectures, which sometimes even
causes bugs (as in the recent case of ATK problem found by Mats). We would
like to discuss and at least internally "standardize" architecture names.
Meanwhile, it cannot be done absolutely uniformly due to some historical
conventions in at least rpm and deb naming. But having them more correlated
is possible. We identified 4 usage types that might need to have differences
in naming of the same architectures:
1. User Level - Navigator, web site, specification, i.e. various human
2. Internal Files - suffixes in various intermediate file names that
LSB tools produce.
3. rpm - suffixes in rpms of various LSB deliverables.
4. deb - suffixes in debs of various LSB deliverables.
And we propose the following names for these types:
The rpm/deb naming is historically established and we do not touch it at
all. The most frequent inconsistency in other areas currently found is using
IA32 vs. x86 and AMD64 vs. x86-64 vs. x86_64 multiplied by low/upper case
variations. We agreed yet almost year ago (the talk initiated by Marc Miller
at the first LF Collab Summit) that it is better to have generic names. Mats
recently mentioned in a comment in bugzilla that LSB 4.0 (as a major
version) might be a good candidate to make it consistent in the
specification. And I just want to advance this issue to make the
"consistency scope" broader.
Do the proposed names look OK to members of this list?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the lsb-discuss