On Fri, 2012-04-06 at 05:26:15 +0200, Guillem Jover wrote: > Hmmm, ok so in 1.16.2 I only fixed part of the issue, I've now fixed the > other part in my local tree, which will be included in dpkg 1.16.3 (to > be uploaded soon). The fix implies looking at the process basename() of > argv[1] instead of argv[0], so it seems --name has never worked on the > Hurd...
Actually no, for --name to work on the Hurd the code will need to check for both argv[0] and argv[1] to handle the cases of binaries and scripts, which makes this all kinds of ugly and even more unreliable than on other systems (where that unreliability comes mostly from the pretty small comm field limit). It would be extremely nice if the command/process name (the traditional comm process field on monolithic kernels) would be tracked by the proc server. I don't remember the exact details now, but it seems to me this might end up touching sourrounding code to the /dev/fd/N issue, so it perhaps could be taken into account while modifying the RPCs? regards, guillem -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/[email protected]
