editors/libreoffice disappeared from ports

drhowarddrfine said:
And as of a few minutes ago, editors/libreoffice is now back as a package and can be installed. The original point of this--and the more current other--thread has been solved and we can move on.
Click to expand...

This is of course still worth exploring. The removal of the package might have been unnecessary and as pointed out it makes us look bad. Especially if it affects quarterly, which could be compared to a stable Debian release (not really, but even reasonable users might do that), which doesn't kill packages.
 
cracauer@ said:
This is of course still worth exploring.
Click to expand...
Perhaps more tractable, I would suggest that pkg could be updated to display unrequested removals of previously requested installs at the bottom of the change list (closer to the prompt, and less likely scrolled off the top) and require an “I UNDERSTAND” or typing the name of one of these removed-for-breakage-on-upgrade packages to prompt users beyond their reflexive ‘y’.
 
cracauer@ said:
This now goes back to the question of why exactly libreoffice failed to build. Was it a bad new version of libreoffice or did some dependency update make it fail although we are still on the old (previously buildable) version of libreoffice.
Click to expand...

We know this. The old version has the same way of failing randomly:

forums.freebsd.org

libreoffice package abandoned

A package upgrade just deleted libreoffice-7.5.5.2_3. It's still in the ports tree, but has enough dependencies to sink the Titanic: [strand.156] $ cd /usr/ports [strand.157] $ grep -i "^libreoffice" INDEX* | cut -c1-70 libreoffice-7.6.2.1|/usr/ports/editors/libreoffice|/usr/local|Full int...
forums.freebsd.org forums.freebsd.org
 
mer said:
Until the next important thing fails to build and gets deleted.....

:)
Click to expand...
It wasn't deleted in the sense that most people think and that term only confuses people. It was only missing so people couldn't install or upgrade. Big difference as I pointed out elsewhere.
I don't see how it makes us look bad either. It was an oopsie for a day or two. No big deal. In my case, it meant I couldn't upgrade my libreoffice for two days. The horror of it all.
 
cracauer@ said:
I don't think it literally failed randomly.
Click to expand...
With all due respect, I don't care what you think. I am in the process of actually looking into the issue, because it happened here a couple of times, and (slowly) it starts to annoy me.
 
PMc said:
With all due respect, I don't care what you think. I am in the process of actually looking into the issue, because it happened here a couple of times, and (slowly) it starts to annoy me.
Click to expand...

So you are saying that it literally fails randomly, as in works and doesn't work in the same revision of the ports tree?
 
drhowarddrfine said:
In my case, it meant I couldn't upgrade my libreoffice for two days.
Click to expand...
But then you couldn't do your power points and spreadsheets and word docs for your status report.

I agree that "deleted" is not an accurate term in this situation. But I think "removed as per user request" while accurate may be met with lots of "I did not do that".
I'd bet that lots of us long term users have run into this a few times and most of us may come here and ask about it and/or do homework by looking at build logs.
I know I have and had the same reaction as you, give a a couple of days and then try.
 
Smells pretty delete-ish to me. I wouldn't blame people for using that term.

If it's truly random with no ports tree changes it is likely caused by the parallel build not being clean.
 
PMc said:
exactly.
Click to expand...
repos are built en masse, using poudrie, I think.
Does the same type of build failure happen if one builds just the single port? I'm assuming "start with a clean tree" in both cases so any kind of dependencies would also get built.
I'm not sure what that shows, but it may be a data point.
 
cracauer@ said:
Smells pretty delete-ish to me. I wouldn't blame people for using that term.

If it's truly random with no ports tree changes it is likely caused by the parallel build not being clean.
Click to expand...

Very probably so. It gets really bad with >16 cores.
 
mer said:
But then you couldn't do your power points and spreadsheets and word docs for your status report.
Click to expand...

Let's make something clear here. This problem did NOT remove libreoffice from my system or anyone else's system. It was perfectly usable throughout the whole term and if you weren't upgrading packages you never would have known there was a problem. Even then, you MIGHT have noticed libreoffice didn't upgrade.
 
drhowarddrfine said:
Let's make something clear here. This problem did NOT remove libreoffice from my system or anyone else's system. It was perfectly usable throughout the whole term and if you weren't upgrading packages you never would have known there was a problem. Even then, you MIGHT have noticed libreoffice didn't upgrade.
Click to expand...
Again, you repeat this for the sake of "clarity" but you are demonstrably incorrect. I (again) tried to explain this in my post here. And a pkg terminal output log proving this was provided here. I have experienced this multiple times with LibreOffice over the years. So have others. That's why we're here. It wasn't a matter of not being able to upgrade to the latest LibreOffice for a few days. It was a matter of LibreOffice being gone entirely from our computers. I ended up using my Windows (which apparently is so broken that it shouldn't be operational) machine to get some work done.

Please don't add to the confusion. Those of us upset that LibreOffice is getting removed aren't imagining things. What we say is happening is indeed happening. I don't know how many other ways to phrase it. Silly semantic arguments re: definitions of "delete" and what constitutes "consent" aside: running "pkg upgrade" is resulting in the active removal of a working package. Can we please move on from suggesting otherwise so that we can move forward versus in circles?

I also take issue with the comparison in this thread of those upset at this being "idiots". I don't think that's necessary or appropriate. There is a valid point to be made here, and I appreciate the attitude of cracauer@. Please don't take the flippant "if you don't like this, go use something else". I've been using FreeBSD since the early 1990s. We're all here because we love FreeBSD and I'm participating in this thread because I see an opportunity for FreeBSD to improve. I'm not a dev... I don't have the means to address this myself. But even as non-coding end-user I should be permitted to take part in discussions. If you see room to educate me more about how FreeBSD works internally, fine... I've already learned a few things from this thread. But don't come storming in here suggesting that myself and others are smoking crack and imagining things. We're a bit smarter than that and aren't "idiots".

I also don't understand why this is such a unique issue with LibreOffice. As far as I'm aware I've never experienced this with any other FreeBSD application.
 
This thread is broken beyond repair. I'm pretty sure any relevant information is already somewhere in here, but buried below tons of rubbish. Any further attempt to "fix" this will only make it worse.
 
The fact that pkg says that it is going to remove a package and asks yes/no is hardly relevant.

If you say "no", then you will prevent security updates for all other packages. An operating system advertising with security can't rely on that.

And secondly, updating from crontabs should be supported without causing massive problems.

I still think a core question is how Debian is dealing with this. It isn't true that they just packport security relevant diffs. Major applications like web browsers and office packages are updated wholesale. Things like desktop environment on the other hand do offer stable branches, and those are used by Debian stable releases. Basically the question is whether upstream offers stable branches. If not then Debian is not backporting security fixes diff-by-diff.

BTW, it is a dirty semi-secret that running Debian-testing means that you get security fixes significantly later.
 
cracauer@ said:
I still think a core question is how Debian is dealing with this. It isn't true that they just packport security relevant diffs. Major applications like web browsers and office packages are updated wholesale.
Click to expand...
It certainly is true. Nowadays, there are (very few!) exceptions to that policy. You might want to read up on why Debian offered some browser called "iceweasel" for quite some time to learn more. Still, the vast majority of (security) fixes for "stable" are done backporting and patching the exact same version. This is a damn lot of work, and thinking it's a better approach would be misleading. See what e.g. rspamd thinks of the "official" Debian packages offered. See what happened to ssh key generation when Debian goofed patching. Well ...
 
zirias@ said:
It certainly is true. Nowadays, there are (very few!) exceptions to that policy. You might want to read up on why Debian offered some browser called "iceweasel" for quite some time to learn more. Still, the vast majority of (security) fixes for "stable" are done backporting and patching the exact same version. This is a damn lot of work, and thinking it's a better approach would be misleading. See what e.g. rspamd thinks of the "official" Debian packages offered. See what happened to ssh key generation when Debian goofed patching. Well ...
Click to expand...

Hmmm, I remember Iceweasel and also that ssh incident.

Maybe I'm mislead by the fact that I use binaries from Google and Mozilla for my browsers on Debian, and I don't use office style software except in the browser.
 
zirias@ said:
So you know what happened? Hm, who told you to report this here? It makes little sense, fallout mail is automatically sent to maintainers (so they're already aware!), and only few ports committers are active on the forums.
Click to expand...
Agreed. Ports developers receive a pkg fallout email when a port fails to build.

Src developers receive a Jenkins email when their commits have caused build failures, though best practice is to build universe on one of the universe build machines.

Submitting a bugzilla bug is the best approach when notifying anyone of bugs. Not here. It's only a minority of developers who frequent forums. And, I might visit once every week or two, maybe even longer. Also, IMO you will get a lot more exposure on the mailing lists than here.
 
cracauer@ said:
If you say "no", then you will prevent security updates for all other packages.
Click to expand...
Solvable by hand - you mustn't proceed a pkg upgrade to get a security update installed. In such cases I've done partially updates which left some ports without security issues behind.
cracauer@ said:
I still think a core question is how Debian is dealing with this.
Click to expand...
It is the same: If you change the environment dpkg / apt-get tries to solve this by deinstalling packages.

But Debian has stable, testing and unstable/sid - and what FreeBSD (quarterly) does matches best testing ("you can get packages that are known to work, but sometimes a dependency of something you've already got installed can be broken"); And so a testing user knows that kind of problem, too. Whereas users using stable ("such problems may occur, but are seldom") blame the old package versions: The price of becoming everything "in one line" means much time at the package freeze (!), and another infrastructure as well as more manpower to the package building environment. And backports, modified instead of original software etc…

But on a closer look it is the same on all package systems with dependencies. Not solvable - except you're freezing all port versions and publish them only if and when everything is solved. Means: Publishing outdated software.

This situation is like you're telling the electrician to update all your sockets to a newer system, but you're still having one device which is at the moment not available with the new plug. On Debian stable all packages have to wait till everything is solved ("freeze"); On FreeBSD you can get the newer versions immediately, but if something is missing you have to say "okay, I'll wait till it is ready" - or you say "okay, the deal of missing XY for some days is okay to me". The default is "no".

Whereby Debian started not to wait until everything is ready - instead of that, nowadays even supposedly unimportant software is simply left out. The schedule has to be kept (in the past: it's ready when it's ready).
cracauer@ said:
Major applications like web browsers
Click to expand...
The procedure for browsers is an explicit exception to the usual Debian procedure.
 
You must log in or register to reply here.
Back
Top