Steve> So the answer is "have the latest semantic and speedbar Steve> installed". Yeah, I guessed that. Steve> Hmm, that's not a very good answer is it? No, it's not. Steve> This is untested, but the theory is "put what we need at Steve> the start of the load-path, so that way it doesn't matter Steve> what's installed". This may be best for non-maintainers, who often will be able to get away with somewhat out-of-date installed packages. The package maintainer should clear the load-path of all installed packages, and always build from a full fresh CVS update (perhaps to a "known good" state rather than current). That way the first line of defense to _all_ bugs is "update all your packages and try to replicate". -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."

Valdis> On Mon, 28 Aug 2000 10:33:25 +0900, "Stephen J. Turnbull" Valdis> said: Valdis> Gaak. Those of us who are SUMO-dependent may just learn Valdis> to live with the bugs. I certainly hope not! I hope such bugs don't make it into SUMO. I was referring to the situation where people are updating random packages between SUMOs. Without a proper dependency checking system, "update all to most recent" is the best we can do. Valdis> I never DID figure out how to get packages to agree with Valdis> 'depot', so I've been living with SUMOs for those. Guess Valdis> it's time for me to deal with the learning curve... ;( ./configure --prefix=/depot/gnu --exec-prefix=/depot/gnu/plat ... has been working for me for just ages. So much so that I've forgotten about it, I just keep doing ./config.status --recheck. There was the package-get breakage (ultimately EFS IIRC?) but that's OK now too. Symptoms? -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."

Valdis> On Mon, 28 Aug 2000 17:37:06 +0900, "Stephen J. Turnbull" Valdis> <turnbull(a)sk.tsukuba.ac.jp&gt; said: Valdis> I still see that as a potential horrorshow - the sysadmin Valdis> installs the most recent SUMO, a user gets his own copy of Valdis> package XYZ, which blows chunks until you discover you Valdis> need to install ABC, byte-compile, install DEF, then XYZ, Valdis> then byte-compile again. I still don't see what to do about that without "truly insane dependency checking." :-( A cure worse than the disease IMO, given that the only real remedy for a failed dependency would be to prohibit installation. Nor would it be really better than what we have now AFAICT; at least once you get package-get running, updating packages is pretty easy. Users who are updating their own suite of packages had better know what they're doing in any environment. If they really can't use what's publically available on the system, they're vulnerable to version skew no matter how you slice it. Valdis> Umm.. many moons ago, when I tried to build packages, I'd Valdis> not be able to convince them they needed to *run* from Valdis> /usr/local/yaddayadda, but that a 'make install' should Valdis> place them in /local/stage/yaddayadda, Oh. I don't see a problem offhand, but I don't do it that way so I'll just leave it at that. -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 _________________ _________________ _________________ _________________ What are those straight lines for? "XEmacs rules."