Subject: Re: Assming char == signed char - example ...
To: None <>
From: Andrew Cagney <>
List: tech-ports
Date: 06/21/1995 12:03:20
[changed mailing list, the topic has changed]

Excerpts from mail: 20-Jun-95 Re: Assming char == signed .. (1302)

> But I'd imagine we'll wait until we're confident we've found most of the
> signed char problems before intergrating back into any m/c independent
> code.

Should the changes be delayed?  I just wonder if perhaphs, in the case
where there is a clear problem with machine independant code (along with
a simple well defined fix) that perhaphs such changes should be merged

I assume that, like anything, it is a trade off.  On the one side, you
don't want #ifdef arm's (say) appearing before they are known to be
correct  and working while on another other side you don't want to delay
by too much fixes to things like `unsigned char ' (1) as otherwize you
could end up with several ports fixing the same problem (a slim chance
but still there :-).  Oh and on a third side there is the added overhead
of having to co-ordinating a multi stage merge....

Just a ramble,



(1) I supose this sys/lib/libkern stuff could possibly come under this


Excerpts from mail: 20-Jun-95 Re: Assming char == signed .. (1302)

>  > (Just don't mention the rugby^D^D^D^D^Dwar ... :-)

> Did we lose then ? ;-)

Hang on.  For you to be able to loose the semi, we had to be `kicked'
out in the quater finals. I  don't remember this ... :-)