***Server sets mode: +nt
<damo22>okay, i can add comments and repush, but it will be later today ***jma is now known as junlingm
<jrtc27>at a glance I can see at least one locking issue <jrtc27>if you drop the lock after seeking, and between each iteration of the while loop, you get race conditions <junlingm>Hmm, somehow, active translators do not work. Does libnetfs support active translators? <junlingm>jrtc27: while thinking about the race condition more, it seems ok, because the file cannot be read from the two different threads. The file is locked for read in netfs. The struct sshfs lock is really to prevent two ops on different files, because libssh does not like it. . <junlingm>anyway, the lock was fixed. better to be safe. <youpi>junlingm: yes, libnetfs supports active translators, etc. procfs uses it <junlingm>youpi: ok. thanks. I will look at procfs to see how to support active translators. <junlingm>youpi: "settrans -a /proc/version /hurd/hello; cat /proc/version" does not seem to produce "hellow world" <youpi>ah, you mean setting an active translator inside the hierarchy of a libnetfs-based translator <youpi>possibly that's not supported indeed <youpi>passive translators would have to involve not only libnetfs, but for active translators it should be possible to add it in libnetfs itself <junlingm>I thought so too. In fact, libnetfs can set the translator, but it cannot use it. So it probably intended to support it.