Systemd Dreams Up New Feature, Makes It Like Cron

[JanC]

You will get the cron-replacement and a bunch more from the systemd git tree. And thats really interesting. If you decide to do like Ubuntu and go all anti-systemd then you also get the pain of maintenance. And the rest of the sane world? They will have a single tree to pull the entire core from.

Now the idiot crowd have to revert to sanity(systemd tree) or sustain the pain and suffering from maintaining crap out of tree. Mark Shuttleworth is probably thinking about ditching upstart too. Any hour spent on core stuff outside the systemd tree is really a wasted effort.

Systemd was started as sort of an anti-upstart (for reasons that probably were a mix of NIH, company/licensing politics and technical disagreements?some of those might have been good reasons), not the other way around. And you do know that integrating cron or a cron-replacement with upstart was on the todo-list of upstart that Lennart used (among other sources) to get ideas for systemd, right?

[DeeK]

Hey, look at that. Looks like it's the Mac & Windows users who are the cheapskates. Linux users? They're willing to pay for quality stuff, when the merchandise is offered on terms that are fair and not covered in EULAs and DRM.

Well, to be brutally honest, it shows that everyone acts cheap. It does make Linux users look less cheap than the rest, though.

[dee.]

Reason: Linux users are whiny bitches who gladly download and use _free_ stuff that other people worked their asses off on and then rarely contribute back anything besides more whining and bitching.

Funny thing, that. So, you claim Linux users are cheap and never want to pay for anything... ok, whatever you say. But hey, let's look at the latest Humble Indie Bundle:

https://www.humblebundle.com/

Average purchase: $6.70
Average Windows: $6.23
Average Mac: $7.95
Average Linux: $10.47

Hey, look at that. Looks like it's the Mac & Windows users who are the cheapskates. Linux users? They're willing to pay for quality stuff, when the merchandise is offered on terms that are fair and not covered in EULAs and DRM. I guess Mac and Windows guys already pay so much for their operating systems they have no more money left for other things... eh?

[johnc]

Reason: Linux users are whiny bitches who gladly download and use _free_ stuff that other people worked their asses off on and then rarely contribute back anything besides more whining and bitching.

There was tons of Linux users whining about Upstart at first. It broke stuff. Then it matured. Same as systemd. Less whining about launchd, but then OSX users don't whine and bitch like Linux users. Nobody does.

All the major systemd complaints are from non-users of systemd. They whine because they can, not because they have any real reason to. Par for the course in Linux land. Constant gobs of "stop energy" and armchair programmers giving their unsolicited, unwelcome, idiotic opinions on what they think good software looks like.

You have never posted a single positive thing about Linux on this forum, and now you're criticizing others?

[elanthis]

I never hear anyone bitch about Ubuntu's Upstart, Apple's launchd, or OpenRC.

People only seem to be bitching about systemd.
Maybe there is a reason?

Reason: Linux users are whiny bitches who gladly download and use _free_ stuff that other people worked their asses off on and then rarely contribute back anything besides more whining and bitching.

There was tons of Linux users whining about Upstart at first. It broke stuff. Then it matured. Same as systemd. Less whining about launchd, but then OSX users don't whine and bitch like Linux users. Nobody does.

All the major systemd complaints are from non-users of systemd. They whine because they can, not because they have any real reason to. Par for the course in Linux land. Constant gobs of "stop energy" and armchair programmers giving their unsolicited, unwelcome, idiotic opinions on what they think good software looks like.

[curaga]

Its 1 time expense of extra work-- the time needed to write the makefile ONE TIME. And if youre really lazy you dont even have to write it once since LFS was nice enough to publish their makefile.

Once per every systemd release, not ONE TIME [sic].

As far as build dependencies... a minimal systemd build (which one would be doing if they just wanted udev) has 3 makedeps. 1) dbus, 2) glibc 3)libcap. All of which would already be installed on the system more than likely so youre issue of "More build dependencies" is BS rigtht there.

Nope. D-bus and libcap have no place on my systems, they are bloat. D-bus is actually absolutely crap in addition, libcap is merely bloat. I suppose people not wanting to use glibc don't exist either.

[Ericg]

So it is only modular if it comes to systemd, not when it comes to compile the udev component, which is still used by many other distribution without systemd.
So those distributions are punished with extra development time (for their own Makefiles) or extra work when maintaining their udev package (longer build time with more build dependencies, the hassle to copy out the udev files) or even the work to fork it, but still people are saying that there is no force towards systemd when you don't want to use it. Man, there were volunteers on the mailing list that wanted to do the work to implement the option to build udev without systemd, but the developers don't want it. But there is no force towards systemd.

Man, your reality must be somewhat distorted.

Its 1 time expense of extra work-- the time needed to write the makefile ONE TIME. And if youre really lazy you dont even have to write it once since LFS was nice enough to publish their makefile.

As far as build dependencies... a minimal systemd build (which one would be doing if they just wanted udev) has 3 makedeps. 1) dbus, 2) glibc 3)libcap. All of which would already be installed on the system more than likely so youre issue of "More build dependencies" is BS rigtht there.

The patches for udev-only? http://comments.gmane.org/gmane.linu...ug.devel/17478 the developers had problems with the patches submitted and the discussion eventually died off. Want the patches? There ya go, resubmit them for inclusion to systemd-dev.

[tomegun]

*sigh* And yet there is still no way to neatly run a scheduled job on the last day of the month. Honestly, if you're going to write a cron replacement, you may as well make improvements to it.

Currently, if I need to do this, I specify the process to run on the 28th - 31st of every month, and include this shell test before the job command: [ `perl -e 'print ((localtime(time+86400))[3])'` -eq 1 ] &&

It works, but I'd much rather have a built in solution to the cron replacement.

Does not sound entirely unreasonable. I'd suggest making a feature request on the systemd mailinglist. I suppose what you want is the posibility of using negative values in timestamps (i.e., the -1'st day of the month, and so on wherever it makes sense). I didn't think about it carefully, but prefixing the value by '-' should work (i.e., I don't think it classes with the range operator which is also '-'), but it might be very confusing to the user, so maybe something better is needed.

[Vim_User]

Its only "there goes the modularity" because systemd has a hard dependency on udev. BUT its not the other way around. Compile systemctl, journactl, and udev together and just copy-paste out udev. Or write your own makefile that only compiles udev.

So it is only modular if it comes to systemd, not when it comes to compile the udev component, which is still used by many other distribution without systemd.
So those distributions are punished with extra development time (for their own Makefiles) or extra work when maintaining their udev package (longer build time with more build dependencies, the hassle to copy out the udev files) or even the work to fork it, but still people are saying that there is no force towards systemd when you don't want to use it. Man, there were volunteers on the mailing list that wanted to do the work to implement the option to build udev without systemd, but the developers don't want it. But there is no force towards systemd.

Man, your reality must be somewhat distorted.

[DeeK]

*sigh* And yet there is still no way to neatly run a scheduled job on the last day of the month. Honestly, if you're going to write a cron replacement, you may as well make improvements to it.

Currently, if I need to do this, I specify the process to run on the 28th - 31st of every month, and include this shell test before the job command: [ `perl -e 'print ((localtime(time+86400))[3])'` -eq 1 ] &&

It works, but I'd much rather have a built in solution to the cron replacement.