> > I think that update-ms-fonts should have some guard against > > re-entrancy. > > I'm not opposed to the idea, but am not implementing it before etch - > the scenario is not that common and I don't want to risk destabilisation > now.
Sure. > Were there any adverse effects of the parallel runs? What happened? I just pressed Ctrl+C immediately, so nothing bad happened. Don't know if any breakage is possible if two scripts will start to unpack the fonts in parrallel at the same time to the same location. However races are always bad... Btw, a simple solution is just not to install user-runnable update-msfonts, but do things in postinst only (either by moving code into postinst, or maybe to some location under /usr/share (not in PATH - so not intended to be manually executed). This way, dpkg-level locking will avoid execution of several instances. And if re-running is ever needed, there is dpkg-reconfigure.
pgpPuhAKOqG8Y.pgp
Description: PGP signature
