Hello Frank! On Thu, Feb 19, 2004 at 12:39:04PM +0100, Frank Küster wrote:
Florian Ernst <florian@uni-hd.de> schrieb:Checking on Woody won't be necessary, since this package will never make it to Woody. Or did I misunderstand you here?Remember backports. Nobody knows when sarge will be released, and since these scripts are so *useful* ;-), it might be that somebody prepares a backport for woody. Still it's not mandatory - if you don't provide versioned dependencies, it's up to the backporter to check. It's just being friendly to backporters.
Right, I didn't think much about Woody after Sarge was released last December[1]... oh, wait. (SCNR) But seriously, I checked the package on current Sarge and Sid and didn't experience any errors so far. The old version 10 of this package was backported and is available at backports.org, but it is missing all the dependencies which cause so much grief... ;) On Woody with the irssi-text backport from backports.org everything worked out of the box now except for libnet-google-perl and libcrypt-cbc-perl which don't exist in stock Woody. The former exists as a backport from Adrian Bunk while the latter wasn't backported so far, thus resulting in two scripts (blowjob and ircsec) with unsatisfied dependencies. The semantics have changed in one dependency (xosd-bin, no backport yet) in a way to render the script osd unusable wih Woody's version, so I've added a versioned dependency on it. Thus on Woody (+backports) 3 out of 222 scripts won't run so far. Unfortunately I also have to admit I've found some more dependencies of single scripts I missed in the first run, now altogether resulting in 32 Suggests:... A new version of this package was uploaded.
Also, at least in general, versioned dependencies might be beneficial if you want to add an optional dependency later. That is, now you have: Depends: foo, and later, since the newly developed fop is so much cooler, you'd rather have: Depends: foo | fop. Then you have to check whether fop really implements yet all of foo's features that are used, and if all fop versions in unstable did it from the beginning. It's easier if you just need to compare changelogs of known versions, instead of testing all possible usages of your program (or all 222 scripts, in your case).
Point taken, I'll keep that in mind. Thanks for taking the time, cheers, Flo [1] http://lists.debian.org/debian-devel-announce-0308/msg00010.html
Attachment:
signature.asc
Description: Digital signature