On Sun, Dec 17, 2000 at 02:29:32AM -0600, W Bauske wrote:
> Consider why there are two C compilers in RH7.0.
> There is a problem somewhere. Otherwise there would
> be only one.
With this it is even well-known where. All familiar with these issues
acknowledge for quite a while that there are spots in Linux kernel
sources which depend too heavy on a particular compiler behaviour.
Those with a less benign disposition are calling them simply "bugs".
Some points are disputed in that if some undocumented corners should
behave that or another way but this does not change the picture
overall. Remember that Linux kernel is not, and never was, written in
Standard C but is some dialect implemented by gcc and this hits from
time to time.
Things are getting cleaned but the matter is delicate and this will
take a while. Besides the situation is not exactly equivalent for
x86 and other architectures; Alpha in particular. OTOH asking from
a distribution maker to put together the whole thing substantially
differently for every architecture would be way too much.
Not that I want to claim that everything is hunky-dory, as it is not,
but it makes me wonder why when there were similar issues in the past
(they were, in times quite a bit more severe than now) there was only a
bit of grumbling here and there and now we have all these noisy
BTW - a separate compiler in a distribution for a kernel is a Debian
"invention" and 'kgcc' name, now picked wider, comes from Conectiva.
Axp-list mailing list
This archive was generated by hypermail version 2a22 on Tue Jan 2 07:48:32 2001 PST
Send any problems or questions about this archive to firstname.lastname@example.org.