Summary: | x11-libs/motif-2.3.3 on x64-freebsd - UilLexAna.c: In function 'yylex': UilLexAna.c:1488: error: request for member 'value' in something not a structure or union | ||
---|---|---|---|
Product: | Gentoo/Alt | Reporter: | matt <mattmatteh> |
Component: | Prefix Support | Assignee: | Gentoo Prefix <prefix> |
Status: | RESOLVED WONTFIX | ||
Severity: | enhancement | CC: | bsd+disabled, ulm |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | FreeBSD | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
matt
2010-10-26 02:22:39 UTC
please update the summary to include the version of openmotif, thx This is not even keyworded on ~x86-freebsd, because you have a nice list of keywords in your make.conf. I could just tell you to go away :). But I am able to reproduce and you attached a patch, so I will add a keyword once it is fixed. Ulrich, do you see any problems with this patch? wrt to ACCEPT_KEYWORDS...yah, scary oops, filed that bug in a hurry, x11-libs/openmotif-2.3.3. x64-freebsd prefix is mostly testing, there is little that is keyworded. ill look at the dependencies and clean up my accept_keywords. thanks. Ulrich, I forgot to cc you...you want to add this patch and ~x86-freebsd and ~x64-freebsd? (In reply to comment #5) > Ulrich, I forgot to cc you...you want to add this patch and ~x86-freebsd > and ~x64-freebsd? Looks like it's the same problem that was already fixed for Darwin and Solaris: <http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/x11-libs/openmotif/files/openmotif-2.3.2-darwin.patch?revision=1.2&view=markup> But why doesn't the issue show up on native ~x86-fbsd then? And, maybe more important: Does the package still compile on ~x86-fbsd with that patch applied? BSD team, could you please test if OpenMotif still compiles for you with the included patch? I'm affraid Prefix differs from vanilla FreeBSD here (and Darwin, and Solaris), because we provide an uptodate GNU (f)lex. Gentoo/FreeBSD should have that too, but IIRC that define is basically a bad way around for a missing configure check. (In reply to comment #8) > I'm affraid Prefix differs from vanilla FreeBSD here (and Darwin, and Solaris), > because we provide an uptodate GNU (f)lex. Gentoo/FreeBSD should have that > too, but IIRC that define is basically a bad way around for a missing configure > check. Ulrich, how responsive is upstream? Is there a way to push the already applied patches and add a proper fix instead of that workaround? (In reply to comment #9) > Ulrich, how responsive is upstream? Responsive enough. > Is there a way to push the already applied patches and add a proper fix > instead of that workaround? Their bug tracker is at <http://bugs.motifzone.net/>, but first we need a proper fix. Two years' ping. Any progress here? The fix depends on version of flex, which in Prefix will always be up-to-date, while usually being odd or out-of-date on non GNU-Linux systems. That said, we're always going to have a mismatch with upstream here. Perhaps, a simpler but more effective approach would be to use: #ifndef YYSTYPE # define YYSTYPE yystype #endif (In reply to comment #12) > Perhaps, a simpler but more effective approach would be to use: > > #ifndef YYSTYPE > # define YYSTYPE yystype > #endif Can we be sure that YYSTYPE is always defined as a preprocessor macro, and not as a union or a typedef? I guess in the general case we can't. The problem obviously lies in the need to define it. I'm sorry, it doesn't seem we're going to fix this stale bug. |