The net-dialup/hsfmodem-7.68.00.04 doesn't behave properly with the init scripts. First, it automatically adds /etc/init.d/hsf to the default runlevel. Second, when unmerged it leaves the hsf script in the default runlevel and doesn't remove it from /etc/init.d/
(In reply to comment #0) > First, it automatically adds /etc/init.d/hsf to the default runlevel. No, it doesn't. In fact, it even doesn't install any initscript at all, it's the hsfconfig that creates/deletes the initscript when you run it with an appropriate option, not the ebuild. > Second, when unmerged it leaves the hsf script in the default runlevel and > doesn't remove it from /etc/init.d/ Nothing removes itself from any runlevel when unmerged, that's an expected behaviour. So, what exactly are you requesting here?
(In reply to comment #1) > (In reply to comment #0) > > First, it automatically adds /etc/init.d/hsf to the default runlevel. > > No, it doesn't. In fact, it even doesn't install any initscript at all, it's > the hsfconfig that creates/deletes the initscript when you run it with an > appropriate option, not the ebuild. I see, I seem to have missed that. But shouldn't the ebuild install the init script instead of the hsfconfig tool? And shouldn't the user manually add it to the runlevel? The ebuild says to run hsfconfig to configure the modem, but the hsfconfig tool doesn't mention the init scripts at all. So it seems to do this in an obscure way, which IMHO is wrong. > > Second, when unmerged it leaves the hsf script in the default runlevel and > > doesn't remove it from /etc/init.d/ > > Nothing removes itself from any runlevel when unmerged, that's an expected > behaviour. > > So, what exactly are you requesting here? As said before, the init script should be installed by the ebuild, and adding it to the default runlevel should be done by the user. Please reopen the bug if you believe this makes some sense.
Shrug; reading the hsfconfig utility, it's a complete mess and definitely a reason why noone bothered to patch it to not create the initscript...
Fixed (sort of) in hsfmodem-7.68.00.04-r1, hcfpcimodem-1.14-r1 and hcfusbmodem-1.14-r1 by running {hsf,hcfpci,hcfusb}config --remove in pkg_prerm when $ROOT==/ and the corespondent init script exists. While I agree it is not desirable to create init scripts like this, I don't have an option here. Unfortunately configure scripts cannot save their work in $D without a serious amount of work. If you are unhappy about it, bug me with a *tested* patch that fixes the issue. Closed as WONTFIX.
(In reply to comment #4) > Fixed (sort of) in hsfmodem-7.68.00.04-r1, hcfpcimodem-1.14-r1 and > hcfusbmodem-1.14-r1 by running {hsf,hcfpci,hcfusb}config --remove in pkg_prerm > when $ROOT==/ and the corespondent init script exists. Well, I've been thinking about this solution but it causes the initscript to be removed on re-emerging the same version as well... not really desired I'd say. I'm afraid there's no good way to do such thing.
hsfconfig do more than simply creating the init script. I would say it is a good thing to force user to re-run hsfconfig everytime he/she installs/update hsfmodem.
(In reply to comment #6) > hsfconfig do more than simply creating the init script. I would say it is a > good thing to force user to re-run hsfconfig everytime he/she installs/update > hsfmodem. Well yeah I did read the script; seriously, all the stuff that's done via hsfconfig should be done in src_compile if this thing wasn't packaged as a horrible home-brewed hack by upstream and was using a sane build system... P.S. Don't get me wrong, this bug is definitely OK as WONTFIX for me ;)