Linked by Thom Holwerda on Fri 30th Apr 2010 18:41 UTC, submitted by diegocg
Thread beginning with comment 421853
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.
His only concrete objections are (1) not all existing daemons would be compatible with launchd, and (2) launchd is not flexible enough to do everything he can conceive.
He also says that launchd is a bit Mac specific.
I believe it would be wiser to use a working, release-quality codebase than to write something from scratch. If more backwards compatibility is needed, it would be much simpler to add this functionality to launchd - there is nothing in its architecture to prevent this.
If systemd had not been written yet, maybe it would have been easier to adapt launchd. However, systemd already works reasonably well, and it might take less work to complete systemd than to adapt launchd to Linux.
Plus, if Linux adopted launchd, daemons could be written to be compatible with both Linux and OS X, and developers from both communities could contribute to the further development of launchd. Both OSs would benefit from sharing an open standard.
I agree. However, the daemon requirements for systemd and launchd are very similar, and it should take little or no work to make daemons work on both systems.
I believe it would be wiser to use a working, release-quality codebase than to write something from scratch. If more backwards compatibility is needed, it would be much simpler to add this functionality to launchd - there is nothing in its architecture to prevent this. And to answer his second objection, sure, launchd is not perfect, but it's much, much better than init/cron/inetd/xinetd/whatever.
To modify launchd, they would either have to fork it or co-operate with Apple. It may be that Linux-specific modifications are not something they want in Apple codebase.




Member since:
2009-12-05
The article points out a few shortcomings in launchd, but admits it does most of what he wants just fine. His only concrete objections are (1) not all existing daemons would be compatible with launchd, and (2) launchd is not flexible enough to do everything he can conceive.
I believe it would be wiser to use a working, release-quality codebase than to write something from scratch. If more backwards compatibility is needed, it would be much simpler to add this functionality to launchd - there is nothing in its architecture to prevent this. And to answer his second objection, sure, launchd is not perfect, but it's much, much better than init/cron/inetd/xinetd/whatever.
Plus, if Linux adopted launchd, daemons could be written to be compatible with both Linux and OS X, and developers from both communities could contribute to the further development of launchd. Both OSs would benefit from sharing an open standard.