Bug 214278

Summary:	rpm/beecrypt uses -fsigned-char
Product:	[openSUSE] openSUSE 10.2	Reporter:	Olaf Hering <ohering>
Component:	Basesystem	Assignee:	Michael Schröder <mls>
Status:	VERIFIED WORKSFORME	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None	CC:	pth
Version:	Alpha 5 plus
Target Milestone:	---
Hardware:	PowerPC
OS:	Linux
Whiteboard:
Found By:	Development	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	rpm.beecrypt-fsigned-char.patch

Description Olaf Hering 2006-10-22 17:52:41 UTC

beecrypt in rpm sources forces -fsigned-char. It probably shouldnt do that, better force 'signed char var' where needed.

Comment 1 Michael Schröder 2006-10-23 10:54:46 UTC

Well, I won't go through the code and fix all bad instances of char. Feel free to send me a well tested patch if you have enough spare time.

Comment 2 Olaf Hering 2006-10-23 11:00:28 UTC

Created attachment 102259 [details]
rpm.beecrypt-fsigned-char.patch

Comment 3 Michael Schröder 2006-11-16 11:55:41 UTC

so ppc uses signed char as default nowadays?

Comment 4 Olaf Hering 2006-11-16 12:38:06 UTC

no it doesnt, thats why it is removed.
-fsigned-char was nothing but a band aid to work around broken code.
look for new 'comparison is always true/false due to limited range of data type' warnings with this patch applied.
char x; if (x < 0);
this has to be 'signed char x'

Comment 5 Michael Schröder 2006-11-16 14:58:42 UTC

Forget it, I don't have time for that. Either you go through all the code, fix every issue, and do all the tests to ensure that it still works or I won't remove the -fsigned-char.

Comment 6 Olaf Hering 2006-11-16 15:34:16 UTC

that doesnt fix the bugs, obviously...

Comment 7 Olaf Hering 2006-11-17 21:52:34 UTC

there are no bugs, and s390x has also no -fsigned-char

the only bug is in lib/query.c on 64bit, u32 cant hold ULONG_MAX

query.c:579: warning: comparison is always false due to limited range of data type
query.c:673: warning: comparison is always false due to limited range of data type

Comment 8 Michael Schröder 2006-11-17 22:04:17 UTC

(query.c is not in beecrypt, but rpm)
If beecrypt works without -fsigned-char, why was it added to the Makefile in the first place?

Comment 9 Olaf Hering 2006-11-19 10:23:51 UTC

no idea why it was added, we wre all a bit clueless back in 2000.

Comment 10 Philipp Thomas 2010-06-17 15:54:56 UTC

After 3 1/2 years can't this be closed, besides that in all this time it never changed to assigned?

Comment 11 Michael Schröder 2010-06-17 15:59:51 UTC

Yes, I think so. CLosed.