Christian Grothoff transcribed 3.8K bytes: > On 05/05/2018 03:01 PM, Nils Gillmann wrote: > >> Agreed, it would be great if you could change the tests to use the > > > > Okay, any idea how the .conf files should be adjusted, given that > > you most likely came up with the config system? > > Actually, we just use some shell-ism for that. > > Regardless, I just added GNUNET_TMP to util.conf, so you should be able > to replace any occurrence of "/tmp/foo" in configuration files with > "$GNUNET_TMP/foo", which will result in $TMPDIR/gnunet/foo (or $TMP, or > /tmp, if $TMPDIR is undefined) being used. > > > I think they don't require absolute filenames, but if we have multiple > > flavors > > of how and where the temporary directory can be located, what do we use in > > place of /tmp/gnunet-something/file then? > > If you need something per-instance (if multiple peers are running), such > as UNIX domain sockets, use $GNUNET_USER_RUNTIME_DIR or for system > services $GNUNET_RUNTIME_DIR. For classical temporary files (say of the > driver of a testcase, we now have $GNUNET_TMP. For per-peer caches use > $GNUNET_CACHE_HOME. All are defined in src/util/util.conf > > > I assume a config-file function that does the same as the function in > > libgnunetutil > > disk.c could be useful. > > The above effectively do the same, right now without the check for > $TEMP, but that could be added if desired.
Nice :) Okay, I have to stop here for today, but I'll work on this soon. If I have more questions I can't solve myself I'll get back to you. Have a nice weekend. _______________________________________________ GNUnet-developers mailing list [email protected] https://lists.gnu.org/mailman/listinfo/gnunet-developers
