Le lundi 23 novembre 2009 à 13:44 -0500, Orcan Ogetbil a écrit :

On Mon, Nov 23, 2009 at 11:54 AM, Dominik 'Rathann' Mierzejewski wrote:

...

On Monday, 23 November 2009 at 17:51, Jerry James wrote: [...]

...

I want a switch that says, "Yes, I know this application uses core fonts. It isn't going to change. Shut up, please."

+1.

Nicholas, I know you love to write scripts that send mass emails, but don't you think this is too much?

FYI I hate it. Only did it because I found no other way to move things forward. Feel free to propose and implement a better way so I can stop mine.

Le lundi 23 novembre 2009 à 20:25 +0100, Andreas Bierfert a écrit :

On Mon, 23 Nov 2009 19:50:19 +0100 Nicolas Mailhot nicolas.mailhot@laposte.net wrote:

...

FYI I hate it. Only did it because I found no other way to move things forward.

Maybe it is time to go for the extreme and help people convert there specs to fulfill the guidelines

In case you didn't notice, this is what happened two weeks ago.

Le lundi 23 novembre 2009 à 13:48 -0600, Chris Adams a écrit :

Once upon a time, Nicolas Mailhot nicolas.mailhot@laposte.net said:

...

Le lundi 23 novembre 2009 à 09:51 -0700, Jerry James a écrit :

...

1. I'm going to nag you forever about a problem you can't fix.

This is false, it can get fixed, either with code changes or by dropping the offending package

Core fonts are not going away, are they?

The infra no, the fonts (or at least part of them) yes

Then why the hate for legacy packages using a legacy interface?

The interface is legacy because it didn't work well, and we do not have users sophisticated enough to undestand their font problems are caused by their wandering in the legacy minefield land some packagers recklessly expose.

Le lundi 23 novembre 2009 à 13:09 -0800, Toshio Kuratomi a écrit :

As notting reminded me yesterday, we do not keep software out of Fedora solely because it is crap. So, although it would be nice to port code that uses core fonts to use fontconfig, it could be counter productive to list them along with other problems discovered in font packages that we can and should fix at the packaging level. You run the risk of getting the script/email that's making these updates ignored even when it's reporting genuine, fixable problems.

I'll do the same offer I made a few years ago. Any of those who package bits that use core fonts step up to write packaging guidelines for core fonts, to do the merge reviews on the associated packages, and generally speaking to become the core fonts guy (or gal) in Fedora, can ask me to whitelist all the core font packages he wants or even to shut down this test completely.

If no one is ready to do that, and is happy to continue to have me do this part because its "fonts" and by default "the fonts sig does fonts", should be happy it costs him only a few annoying mails that state I don't like it one bit.

On Mon, 2009-11-23 at 19:48 +0100, Nicolas Mailhot wrote:

Le lundi 23 novembre 2009 à 09:51 -0700, Jerry James a écrit :

...

1. I'm going to nag you forever about a problem you can't fix.

This is false, it can get fixed, either with code changes or by dropping the offending package

Maintainers do not equal developers. I am a package maintainer, yet my coding skills extend to copying and pasting things from Google results. usually incorrectly. You cannot assume that all the people to whom you are sending these mails have the capability to fix code problems, that is not a valid assumption.

Nicolas Mailhot nicolas.mailhot@laposte.net writes:

I don't assume they have the capability to fix code problems. I assume they're ready to do their maintainer work and nag upstream till it does. Or are ready to drop the package if it costs them more than it's worth.

You are assuming that upstream will agree that it's a problem.

I don't have a horse in this race, since none of my packages contain any fonts. I will note though that my own response would involve procmail'ing these complaints to /dev/null.

regards, tom lane

Am 23.11.2009 22:50, schrieb Adam Williamson:

He already said that he had talked to his upstreams and they had said they would not adjust their code. In that case, he really has done everything he possibly can in his position as maintainer, and sending him further nag emails is achieving nothing constructive and serving only to annoy him.

And eventually may lead to packagers stopping to contribute to the Fedora Project. I do not wish to be used as a medium for a fight between a stubborn Font Guideline Evangelist and a stubborn upstream. If it's upstream's decision not to adjust their code then that's the way it is. If the package does not cause any unexpected behavior by not following the Font Guidelines there is nothing further I as a package maintainer am required to do.

Felix

Once upon a time, Nicolas Mailhot nicolas.mailhot@laposte.net said:

To repeat myself once again, core fonts are not free, they have a maintenance cost,

What is the real maintenance cost?

You have said that core fonts are not going away, so the maintenance cost will not go away. How is badgering other maintainers a good thing?

If you don't want to maintain something, then the normal way is to orphan it and let someone else take the job, not badger everybody else using the thing you don't want to maintain anymore.

On Tue, Nov 24, 2009 at 04:46:32PM +0100, Jochen Schmitt wrote:

If you mean the original bitmap oriented fonts of X11, so they are several reasons to avoid the usage of this kinds of fonts.

The may issue with this fonts is, that they are not scaleable to any size you want.

So what - perhaps the application doesn't need to scale the fonts to any size? Perhaps the application doesn't need to use anything other than 'fixed'?

Just because Xft fonts are better does not mean core fonts are useless.

Rich.

Le Mar 24 novembre 2009 16:00, Chris Adams a écrit :

Once upon a time, Nicolas Mailhot nicolas.mailhot@laposte.net said:

...

To repeat myself once again, core fonts are not free, they have a maintenance cost,

What is the real maintenance cost?

You have said that core fonts are not going away, so the maintenance cost will not go away.

The costs could go down to nothing if there was no core font user left in Fedora

How is badgering other maintainers a good thing?

It is reducing the problem envelope.

If you don't want to maintain something, then the normal way is to orphan it and let someone else take the job, not badger everybody else using the thing you don't want to maintain anymore.

Does not work that way. If it was a clear package dependency, I could orphan the stuff, and all the people who complain at me now would be forced to take themselves in charge and do the work needed by the stuff they use. Because of the brain-damaged way core fonts were specified, the dependency is not expressed in that way and I can not stop caring about core fonts without stopping caring about other fonts (because as long as I have a fonts hat, and no one has a core fonts one, people come to me by default and don't want to hear about differences in font systems).

If you insist, I can ask formally FESCO if it thinks I do more harm than good. It it thinks so, I'll do the logical thing, and go back to being just the DejaVu maintainer, which was actually fun to do, and less time-wasting besides.

Le Mar 24 novembre 2009 17:01, Chris Adams a écrit :

That's not an answer. What is the real maintenance cost?

I already explained yesterday : there are rotting Fedora Core packages to merge review, packaging guidelines to write to define how they are supposed to be cleaned up, a huge pile of existing fonts to re-check for licensing, a huge pile of fonts to re-check for technical soundness (ie a lot of fonts for that area are not encoded properly or declare bad names, should it continue to be hidden via manual fonts.dir or should they be converted to something cleaner, it we continue to go the manual fonts.dir way someone needs to review existing files) etc.

We still had crashes this year due to problems in some fonts.dir.

When i18n asked what was the exact need for bitmap-fonts no one answered. There is a need of someone that can answer other Fedora groups when such questions are asked.

etc, etc

Le mardi 24 novembre 2009 à 10:44 -0600, Chris Adams a écrit :

Once upon a time, Nicolas Mailhot nicolas.mailhot@laposte.net said:

...

Le Mar 24 novembre 2009 17:01, Chris Adams a écrit :

...

That's not an answer. What is the real maintenance cost?

I already explained yesterday : there are rotting Fedora Core packages to merge review, packaging guidelines to write to define how they are supposed to be cleaned up, a huge pile of existing fonts to re-check for licensing, a huge pile of fonts to re-check for technical soundness (ie a lot of fonts for that area are not encoded properly or declare bad names, should it continue to be hidden via manual fonts.dir or should they be converted to something cleaner, it we continue to go the manual fonts.dir way someone needs to review existing files) etc.

And how much of this is still going to be done no matter what, since core font support is not going to be dropped?

You confuse core font support (=xorg code) and core fonts (= rotting font files that no one wants to maintain, and the associated fonts.dir indexes that break regularly)

To keep core fonts support available anything but the built-in fonts (not the full historic xorg font suite, just the single fallback font built in xorg) can be dropped.

Of course the packagers of the apps that use this stuff are going to howl, since it will reduce what their apps can do, but none of them have shown the slightest interest in contributing to the maintenance of the stuff they use so far (to keep things interesting a lot of said apps request fonts without checking they are actually available, and will crash if those fonts are not present. But that's not a core fonts support problem, that's a coding problem in those apps. They can be broken without technically removing core fonts support from Fedora).

In fact one conclusion of this thread is that core fonts users are emphatically not interested in contributing to the stuff they use, and that it's better to remove the most rotten parts from Fedora, instead of keeping it, and continuing to wait for them to fix it.

Like I said, this can be done without removing the xorg code Fedora is commited to maintain to keep X11 protocol compatibility.

Le lundi 30 novembre 2009 à 19:24 -0600, Matthew Woehlke a écrit :

Nicolas Mailhot wrote:

...

When i18n asked what was the exact need for bitmap-fonts no one

answered.

Bitmap-fonts is an exact package name installed by default for no reason anyone would justify

Legibility?

I don't know about font systems, is Terminuis a "core font"? It is bitmapped, but I don't know if that automatically makes it a "core font".

It does not make it automatically a "core font".

Look, people, either take my word core fonts are bad, and induce maintenance, or make the effort to document yourself.

Le Mar 24 novembre 2009 17:06, Jeremy Sanders a écrit :

Nicolas Mailhot wrote:

...

The costs could go down to nothing if there was no core font user left in Fedora

Surely some are required for external legacy applications (including free software and propitiatory applications)?

If no one Fedora-side wants to do the work, this is a problem for RHEL. The Fedora xorg team as far as I've seen only checks the core fonts code stays operational and the built-in core fonts work. Current Fedora core fonts apps exercise much more than that.

Le Mar 24 novembre 2009 17:09, Jussi Lehtola a écrit : uestions on the internet.

Instead of ranting about legacy fonts that have been used for decades, you can direct your energy towards something useful: making sure that new fonts that are compatible with modern font handling systems are correctly packaged.

As I've already stated, I'm ready to ask the usefulness question to FESCO. But I don't owe you doing one and not the other if you don't want to help me so the other does not interferes with my main interest.

Le mercredi 25 novembre 2009 à 01:11 +0100, Matěj Cepl a écrit :

I think you didn't get the message ... you are very welcome (well, I guess, you are) to fix packaging of those fonts (see bugs at http://is.gd/52X4m). If you say, that somebody else should do the work (I was looking at it whether I could help there, but my estimate was something like a week of full time work), then let that somebody else to decide whether it makes sense.

Thanks Matěj

To sum up the situation:

1. The long-deprecated core fonts ecosystem is composed of users, some code xorg side, and a large pile of font files this code accesses

2. The xorg code at at least one built-in backup font to use it are not going away

3. The large pile of core fonts, OTOH, is in bitrotting packages that didn't see real maintenance for a long time (were bitrotting way before the Fedora Core handover and were never cleaned up since). They still regularly fail and cause crashes that make a bad rep to Fedora fonts.

4. Some people like me and Matěj have been spending or were planning to spend some time trying to fix this. Not because we use this stuff, but because of some misplaced sense of duty. It is not fun stuff at all, however, and anything that helps reducing the enveloppe to fix is welcome news.

5. The real users of this stuff never contributed a bit to this maintenance, avoid answering questions when people ask something about it, refused to write packaging guidelines to help others do this work for them when (repeatedly) invited to, and react in a very hostile manner when they get a single mail asking them to make some effort to stop using this stuff (either patching it out, convincing their upstream to do this change, or finding another non-core-fonts-using alternative to package in Fedora, there are many possible solutions). They were *not* asked to help cleaning up the font packages themselves, because, after all this years of no action, it's pretty clear none of them want to.

In other words, they collectively expect someone else to do the ugly, boring, and time-consuming work on the stuff they use, and BTW this someone else should shut up about it and not remind them they behave like parasites (let's call things by their real name).

Given all this, I'm not motivated a lot to contribute to this work anymore (if I end up doing it it's because I promised people like Matěj help, not because I feel like helping core fonts users anymore. If Matěj and others do their part I'll help them but that's all). So I suppose the only remaining course of action is to:

A. investigate if it's possible for a core fonts client to make sure it only accesses the built-in backup core font.

B. if not investigate it it's possible to write a small proxy lib with a core-fonts-like api that do not let an app select any font but the built-in backup core font

C. if A. or B. are true, orphan all the core font packages in Fedora, and give their current users the choice of : a. drop their core font use b. make sure they only access the built-in backup core font (possibly, writing B themselves) c. actually pick up the maintenance of the core fonts packages they use, starting with writing (and getting approved) clear guidelines how they should be packaged, then passing each of the core fonts packages they need through a thorough Fedora review (not let anyone re-import them in their current awful state, then sit up doing nothing about it)

In others words, go through the orphan route that was suggested by others in this thread.

(RHEL will of course make its own choices)

Dne 25.11.2009 10:03, Nicolas Mailhot napsal(a):

If Matěj and others do their part I'll help them but that's all). So I suppose the only remaining course of action is to:

When we are clear now, I want to be clear as well ... IMHO, fixing these bugs (http://is.gd/52X4m) is at least a week of full-time work for me (it probably includes some time learning the craft, so somebody else might do it faster). Given my current work preferences (especially love I get every day from abrt bugs), I don't expect to do anything about these bugs at least this year.

Matěj

2009/11/25 Nicolas Mailhot nicolas.mailhot@laposte.net:

5. The real users of this stuff never contributed a bit to this

maintenance, avoid answering questions when people ask something about it, refused to write packaging guidelines to help others do this work for them when (repeatedly) invited to, and react in a very hostile manner when they get a single mail asking them to make some effort to stop using this stuff (either patching it out, convincing their upstream to do this change, or finding another non-core-fonts-using alternative to package in Fedora, there are many possible solutions). They were *not* asked to help cleaning up the font packages themselves, because, after all this years of no action, it's pretty clear none of them want to.

I never intended to direct any hostility at you, Matěj, or any other person. I apologize for giving such an impression. I *did* intend to direct hostility at the script that mails out the nag messages. :-)

But my reaction wasn't just to a single email. I have felt the noose tightening around my neck for some time now. I have posted to this list before about my woes as a package maintainer, and so to be nagged about what I have already described as the limitations I am forced to work within upset me. I think both of us feel backed into a corner we don't want to be in. That's bad, because people tend to become aggressively defensive in such situations. I've just dunked my head in a bucket of cold water to cool myself off, so let me try to give a dispassionate account of where I'm standing. Perhaps if we understand each other's positions better, we can find some way to get what we both want.

For most of my upstreams, dropping to a single fallback core font is actually okay, because they are only using fonts at all to pop up an X window showing what they would otherwise display on the console. A single fixed-width font meets that need.

My big problem is with XEmacs. It uses a variety of core fonts for different parts of its display. The display engine itself is very old code that no current developer understands thoroughly. An effort was made a few years ago to port our code to fontconfig + Xft. It worked just well enough that some people want it turned on by default, but it actually still has numerous problems (e.g., what the Release Manager calls "display turds", where a few pixels remain after a character is supposedly erased). Meanwhile, the developers who started that effort went on to other things and left the code in its unfinished state. No current XEmacs developer really has a good handle on that code, or what needs to be done to finish it. Therefore, I feel forced to continue building XEmacs in Fedora with the old core font code.

The options I have been offered are, as I understand it, to:

1. Step up to help out with the maintenance of the core fonts. I don't know anything about any font system, legacy or modern, and precious little about X. I don't have even the faintest glimmer of a clue about how to start doing this.

2. Nag my upstream to fix their application. Well, I AM upstream, and no current developer knows how to finish this project. We would gladly accept help.

3. Fix the application myself. I don't know how to do this. My expertise lies in low-level systems work. I don't know much of anything about GUIs in general.

4. Orphan the package. As an upstream developer, this is my baby we're talking about. I would be VERY unhappy if XEmacs were to be dropped from Fedora. (Not that I expect anyone else to be concerned about my happiness; I'm just saying that I'm not very willing to do this.)

I don't have a good option here. Given sufficient time, I believe we can stumble our way to a working fontconfig + Xft setup, and then it won't be a problem, but right now I can't really do anything about the situation. How long do we have before the core font packages are removed from Fedora? A month? A year? Two years?

In other words, they collectively expect someone else to do the ugly, boring, and time-consuming work on the stuff they use, and BTW this someone else should shut up about it and not remind them they behave like parasites (let's call things by their real name).

I believe that you just described what every package user expects of a package maintainer. That doesn't warrant calling the package users "parasites".