"Kevin B. McCarty" <[EMAIL PROTECTED]> writes: > Brian Nelson wrote: > >> It's not clear to me that splitting out the headers is actually a good >> thing (it's very annoying for autobuilders since the corresponding -dev >> package won't be installable until the new version has been autobuilt), >> so I certainly don't think policy should endorse it. > > It wouldn't be an endorsement, just a permission. It seems to me that > policy currently prohibits -headers packages for shared libraries by > saying that development files must be in the -dev package. Do you feel > -headers packages _should_ be explicitly prohibited?
Sure, I don't see any advantage to having them. > My main motive in making the suggestion is that when the headers are > architecture-independent and there are a lot of them, splitting them out > into a separate arch:all package can save a lot of archive space. (I > don't know what the motive was of the developers who created packages > like libqt3-headers, which are arch:any.) Get:1 http://rubeus sid/main libqt3-headers 3:3.3.5-3 [364kB] Fetched 364kB in 1s (226kB/s) I wouldn't call 364kB a lot of saved archive space, and you'd be hard-pressed to find a package with more headers than Qt. >> "Kevin B. McCarty" <[EMAIL PROTECTED]> writes: >>>CC'ed to debian-devel in case anyone wants to add to or disagree with >>>this suggestion. >> >> Uh, no it's not. > > For "CC'ed" read "X-Debbugs-CC'ed". The web archive of the mailing list > hasn't been updated since early this morning (as of this writing), but > you can see my email in the gated newsgroup, for instance on Google > Groups, http://groups.google.com/group/linux.debian.devel . > If it isn't showing up to debian-devel email subscribers, something > strange is going on (I read the list through the web archive so I don't > know whether or not this is the case). Oh, I guess the mail I replied to hadn't been processed by the submit bot then. My mistake. -- Captain Logic is not steering this tugboat. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
