Hi Oliver, *, On Mon, Apr 11, 2005 at 09:32:05AM +0200, Oliver Braun wrote: > [ rpm and ConcurrentAccess ] > The chosen approach to create a scriptlet in /tmp from the postinstall > script does also have the disadvantage to break when using the --root > parameter during install. > > Since I have noticed that at least the rpm version which does not block > the concurrent access for me does have the following feature build in: > [...] > I wonder > > a) if the observed behavior is directly related to this feature and
You mean sometimes rpm waits and sometimes it returns immediately with an error-message? No, this doesn't relate to the feature. If it the version of rpm had the feature, it would not need to wait to give the result. > b) if an rpm feature can be queried from a postinstall scriptlet, only with ConcurrentAccess available. The bad thing is that when you try to access the database and rpm doesn't support concurrentAccess, rpm may lock up and has to be killed. > so > that we could stop writing a scriptlet in /tmp on systems where this is > not necessary. The only thing I could imagine is creating seperate packages, one with "Requires: rpmlib(ConcurrentAccess)" that doesn't need the external scriplet and one "legacy" package that uses the delayed/detached execution with the script in /tmp > Any help on this greatly appreciated. Another idea that came to my mind is something like "%trigger -- rpmlib(ConcurrentAccess) AND openofficeorg-core01" but I doubt you can link the triggers with AND :-( So different packages probably are the only solution. ciao Christian -- NP: 4Lyn - No 11 --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
