mån 2003-03-24 klockan 10.48 skrev Herbert Xu: > On Sun, Mar 23, 2003 at 10:45:20PM +0100, Martin Sj?gren wrote: > > > > It's harder to parse. I don't want anna to screw up because someone > > named their udeb a bit weird. The pattern for kernel module udebs so far > > seem to be on the form *-modules-*.*.*-*-udeb which is a bit too unclear > > for my taste :/ > > This seems pretty simple to me. Find "-modules-" in the string, > and then keep going until you see "-udeb" at the end.
The parsing was the least concern. Bad explanations on my part, I'm worried about false negatives and false positives. Most udebs end in "-udeb" so that's far from conclusive. And I can imagine udebs with "-modules-" in the name, even if we have none now. I guess we should cross-run the info in the name with the version of the udeb to be on the safe side. I have, however implemented the first step (parsing the package name) and the CVS version seems to work fine. /Martin
signature.asc
Description: Detta =?ISO-8859-1?Q?=E4r?= en digitalt signeradmeddelandedel