Hans-Christoph Steiner wrote:
Hey IOhannes,
http://pure-data.svn.sourceforge.net/viewvc/pure-data?view=rev&revision=...
Just checking out your commits. The ./pd-gui.tcl [[<host>:]<port>]
thanks
stuff looks interesting, but your commit removed that stringent test,
i hope it did not remove any stringent text. the patch will only take effect, if there is _exactly_ a single argument and this argument is either - exactly an integer (<port>) - or exactly a something followed by a ":" followed by an integer
so the only relaxation of tests i did was: - remove the "5400" barrier for ports (what was that for?) - additionally allow the number to be prefixed by a word ending with a colon.
which might be needed if "wish pd-gui.tcl" is going to be able to handle command line args like -stderr -open -verbose, etc.
i have tested (see bug #2929975) and things like "wish ./pd-gui.tcl -stderr delme.pd" still works as expected. also "./pd-gui.tcl delme.pd" does what it should.
you cannot open patches named "localhost:5400" (in theory you should, by using the "-open localhost:5400" syntax; but it seems like this triggers an unrelated check (for .pd suffix?) which prevents opening a so-named file)
Just something to test and look out for.
sure.
Also, what are the security implecations of this?
hmm, i think the implication is, that pd-gui can now be used to lauch a DoS attack, simply by starting a lot of pd-gui's connecting to a remote host. apart from that, it brings back functionality lost during the rewrite: the old Pd-Gui could be started as ./pd-gui 5600 remotehost which would connect to remotehost:5600
the new syntax "./pd-gui remotehost:5600" is afaict more readable to anybody who has ever worked with a networking app on the console.
i "heavily" use (read: rely on) this feature for peerdata.
fgamsr IOhannes