Roland Mainz <[EMAIL PROTECTED]> wrote: > Joerg Schilling wrote: > > Roland Mainz <[EMAIL PROTECTED]> wrote: > > > Yes, and the XATTR API seems to be incompatible to the POSIX shell. > > > Wonderfull... ;-( > > > > Let me give a short answer in advance: > > > > If you are able to prove that the Sun XATTE interface is incompatible > > with this text: > > http://www.opengroup.org/onlinepubs/009695399/utilities/sh.html > > Umpf... I guess I can't do that ad-hoc... CC:'ing ast-users for help...
Well the problems you did name, seem to be a result of the specific ksh93 implementation rather than caused by a POSIX incompatibility of the XATTR interface. > BTW: In the original proposal it would no longer be possible to handle > filesystem objects with the name "-@", right ? :-) This is not a problem of the XATTR interface. Also note that it is obvious that you could either call cd -@ or cd -@ dir where in the first case '-@' is the directory or file name. > But: This is not the problem. The issue is LARGER than the korn shell or > any shell. It is a fundamental implementation issue which affects all > applications and should be addressed in a way which fixes the problems > for all applications (and not via the % cd -@ # hacks). I don't see problems in a typical shell. > > let us discuss the Sun XATTE interface. > > > > If you however are unable to prove that, let us fix the korn shell and then > > add the "rejected" interface. > > Well, it's not me you have to convince - you have to provide a working > patch (the original one didn't work as described previously) and > convince upstream to take it. Given the complexity and amount of work > required this will happen after ksh93s is out. If upstream accepts it > (which is VERY unlikely given the previous comments from David and Glenn > (like > https://mailman.research.att.com/pipermail/ast-users/2006q2/000928.html)) > we will have to support it, but I don't waste time with such a work > unless some fundamental issues with the XATTR implementation in Solaris > have been discussed and worked-out. I don't have the time to do that > IMO there should be an easier solution which applies to ALL shells and > appications instead of rewriting larger parts of ksh93 just to fit the > current XATTR implementation in Solaris. If the current ksh93 implementation suffers from implementation problems, this is a ksh93 problem that should not be used to avoid an unrelated other interface. The Sun XATTR implementation is clean and it is the only XATTR implementation that may act as an umbrella for all other (limited) XATTR implementaions found on other OS. Note that during the past 5.5 years nobody did complain about this interface. If you have a better proposal than the current Sun XATTR interface provides, fell free to explain it to us. Also note that I see no need to make XATTR objects visible objects in the standard FS namespace (*). The solution to access these objects in the Sun XATTR implementation is to chdir into the parent object and then use relative path names. This method allows shells to access XATTR objects. *) There are no visible 'ACL objects' either.....and XATTRs are just another method to extend the file meta data. Jörg -- EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin [EMAIL PROTECTED] (uni) [EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
