(Fedora 35)
In the original post of the "font longevity questions/", I said:
By the way, why do so many fonts show up twice in the font selection tools?
First, the only fonts I recall installing myself are the "Nimbus Roman No9 L" fonts. The tools are (in gnome) "Fonts" and "MATE Font Viewer". There might be others not showing up when I click gnome's "Activities".
I did two screen captures to show what I'm referring to. They're on the google drive here: 1. "https://drive.google.com/file/d/1BkB55RW4jqym7_8xRdkExPtPD_XxR1Np/view?usp=s...". 2. "https://drive.google.com/file/d/1b7Pe_ASGEkfJj8XnIZAVaoznPHj1UOsp/view?usp=s...". Notice in the first, each of the "Nimbus" fonts shows up twice. When I clicked on them, I saw no difference. In the second screen capture, some "Open Sans" fonts show up twice. When I clicked on them, I saw no difference. Is this how it should be, or is there a bug in the font tools, or are some fonts actually on my work station twice?
A second issue is that fonts that are different show up in the font tools with the same name. In this screen capture (on the google drive): "https://drive.google.com/file/d/1qMfya_ObKot3jNbzYjTI7y17Kb0S83lz/view?usp=s...", notice 4 fonts called "Noto Sans". They are different from each other. Notice 4 fonts called "Noto Sans, Bold". They are different from each other. There are other examples. This is only my opinion, but they should have different names. Is the problem in the font tools or in the fonts themselves? If the problem is in the fonts themselves, then there should be a standard requiring unique naming of fonts.
On Mon, 2022-05-23 at 14:05 -0600, home user wrote:
are some fonts actually on my work station twice?
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1 /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.afm /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1 /usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic
A second issue is that fonts that are different show up in the font tools with the same name. In this screen capture (on the google drive): "https://drive.google.com/file/d/1qMfya_ObKot3jNbzYjTI7y17Kb0S83lz/view?usp=s...", notice 4 fonts called "Noto Sans". They are different from each other. Notice 4 fonts called "Noto Sans, Bold". They are different from each other. There are other examples.
For things like that, the first thought that springs to mind is someone, or several people, have created an inaccurate copy of a font (for some older traditional fonts, that often seems to happen). The next thing is whether the font has been created in multiple languages and they're being separately considered.
For what it's worth, I've often noticed that Roman characters in Asian fonts are often diabolical. The sizing and spacing is all over the place. That's not unique to Linux, I'd often see that kind of printing in service manuals, where the English text looked like it's been printed using a 100 year old typewriter that fell off a cliff. And I get that in PDF files, from time to time, where I don't have the font they used, they didn't embed it, and the automatically picked font uses characters with oddball sizing, spacing, some overlapping, like a ransom note.
On 5/23/22 10:09 PM, Tim via users wrote:
On Mon, 2022-05-23 at 14:05 -0600, home user wrote:
are some fonts actually on my work station twice?
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1 /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.afm /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1 /usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic [... snip ...]
This has me believing that we need (changes to?) policies and standards. Those "different things" should not add a font unless... 1. the font is not already installed; .OR. 2. the font already on the system is a proper subset of the font that the "different thing" wants to install; .OR. 3. the font that the "different thing" wants to install is fixing glyphs documented as broken in the font already on the system. If #2 and/or #3 is the case, then the "different thing" should not replace the old version of the font with its version unless the sys.admin. clearly, specifically, directly approves.
Question: How do I propose/submit a policy addition or change? (presumably not by submitting a bug!)
On 5/24/22 1:28 PM, home user wrote:
On 5/23/22 10:09 PM, Tim via users wrote:
On Mon, 2022-05-23 at 14:05 -0600, home user wrote:
are some fonts actually on my work station twice?
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1 /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.afm /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1 /usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic [... snip ...]
This has me believing that we need (changes to?) policies and standards. Those "different things" should not add a font unless...
- the font is not already installed;
.OR. 2. the font already on the system is a proper subset of the font that the "different thing" wants to install; .OR. 3. the font that the "different thing" wants to install is fixing glyphs documented as broken in the font already on the system. If #2 and/or #3 is the case, then the "different thing" should not replace the old version of the font with its version unless the sys.admin. clearly, specifically, directly approves.
Question: How do I propose/submit a policy addition or change? (presumably not by submitting a bug!)
George's later post to this thread: "Unfortunately, widely used open source fonts often get multiple forks [etc.]" has me realizing that the possibility of multiple-branch forks in fonts could happen. So my 1 .OR. 2 .OR. 3 above is too simple. For this branch of this thread, I'm now out of good ideas.
On 5/23/22 21:09, Tim via users wrote:
On Mon, 2022-05-23 at 14:05 -0600, home user wrote:
are some fonts actually on my work station twice?
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1
These are different font formats for older applications that can't read truetype. The only one that should show up in the system fonts is the .otf file.
/usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.afm /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1
This is a different face than the previous one.
/usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic
This is a private directory for GhostScript, not a system font path.
There's nothing here that would cause duplicate fonts to show up.
Tim: [re duplicates of some fonts]
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1
Samuel Sieb:
These are different font formats for older applications that can't read truetype. The only one that should show up in the system fonts is the .otf file.
That may be. You'd kind of hope that applications get fonts through a handler in the middle that comes up with a sensible one-of-everything list.
Though does it preclude a font browser from finding all of them? It has a bit of a different purpose than the font list in a word processor.
/usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic
This is a private directory for GhostScript, not a system font path.
There's nothing here that would cause duplicate fonts to show up.
That rather depends on what you used to explore what fonts you have on a system, surely? Since ghostscript is a well known thing, it's not beyond imagination that a font finder might look there.
If I look at the font viewer on an older Fedora mate installation, it certainly does find some fonts more than once. Rather ironically, a mass of DejaVu fonts, for instance, amongst several other font houses; Nimbus, Noto, & URW. And in the few that I checked, it was listing open type and postscript fonts of the same typeface as two separate things. I'm not sure how a font viewer should decide that
NimbusMonoPS-Bold.afm NimbusMonoPS-Bold.otf NimbusMonoPS-Bold.t1
are identical fonts, for instance, unless there's some meta data in each of them that says so. There's bound to be a variety of software that takes the simplistic view that different filenames mean individual fonts, and if they can handle the different formats they may as well list them.
Unfortunately it requires individual digging in the font viewer to see that identically named "URW Gothic, Demi" and "URQ Gothic, Demi" are open type and postscript, rather than it being immediately obvious. In some programs it was useful to know that your on-screen font had a twin postscript font that would print nicely, and look the same, on your expensive printer, and some word processors gave indications about that. But I think we've got to the stage where we didn't need to do it any more, one font format ought to be good on screen and page.
I also seem to recall some very old software doing that (double listing) with different styles of a font. e.g. If you had example, example-bold, example-italic font files, they'd get listed individually, as well as the base example being re-rendered as bolded, and example being re-rendered italicised.
On 5/24/22 19:39, Tim via users wrote:
Tim: [re duplicates of some fonts]
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1
Samuel Sieb:
These are different font formats for older applications that can't read truetype. The only one that should show up in the system fonts is the .otf file.
That may be. You'd kind of hope that applications get fonts through a handler in the middle that comes up with a sensible one-of-everything list.
Though does it preclude a font browser from finding all of them? It has a bit of a different purpose than the font list in a word processor.
Newer applications use fontconfig, which is the central font management system. Checking on my system, the fonts app shows me 3 "Nimbus Mono PS, Regular" fonts. There's an info button on the top which tells you which font file it's getting the data from. The 3 files are: /usr/share/fonts/lilypond/NimbusMonoPS-Regular.otf
/usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf
/usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1
So, I was not entirely correct earlier. It will take both the opentype and the type 1 fonts and doesn't de-duplicate them which is unfortunate.
/usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic
This is a private directory for GhostScript, not a system font path.
There's nothing here that would cause duplicate fonts to show up.
That rather depends on what you used to explore what fonts you have on a system, surely? Since ghostscript is a well known thing, it's not beyond imagination that a font finder might look there.
A font viewer should only use the system font folders and any modern font viewer will use fontconfig, which won't be looking in random directories.
On Tue, 2022-05-24 at 22:14 -0700, Samuel Sieb wrote:
A font viewer should only use the system font folders and any modern font viewer will use fontconfig, which won't be looking in random directories.
It's not really random directories, though. There's the everyone's fonts in a system location, there may be ghostscripts fonts in a well known location, and there may be a user's fonts in their homespace.
On Wed, May 25, 2022 at 2:14 AM Samuel Sieb samuel@sieb.net wrote:
A font viewer should only use the system font folders and any modern font viewer will use fontconfig, which won't be looking in random directories.
The fontconfig system has a well defined way to configure the directories it searches, see "man 5 fonts.conf". I would not be surprised to find that some applications add "random" entries to the fonts.conf directories. There are also ways to specify a user font directory. Some linux distros provide this "out of the iso".
The font conf files can define "aliases" so that a document that requests a given font may use a "similar" substitute. This generally works for legacy documents in Western languages, but there have been instances of people adding new glyphs to support for other languages without changing the name of the font.
On Tue, May 24, 2022 at 1:18 AM Tim via users users@lists.fedoraproject.org wrote:
On Mon, 2022-05-23 at 14:05 -0600, home user wrote:
are some fonts actually on my work station twice?
That can happen. Different things may provide those fonts.
[tim@rocky ~]$ locate NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.afm /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1 /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.afm /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1 /usr/share/ghostscript/Resource/Font/NimbusMonoPS-Bold /usr/share/ghostscript/Resource/Font/NimbusMonoPS-BoldItalic
Fonts are managed by the fontconfig library. Locate finds auxiliary font metric and encoding files. The proper tool is "fc-list":
% fc-list NimbusMonoPS-Bold /usr/share/fonts/urw-base35/NimbusMonoPS-Italic.otf: Nimbus Mono PS:style=Italic /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.otf: Nimbus Mono PS:style=Bold /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.otf: Nimbus Mono PS:style=Bold Italic /usr/share/fonts/urw-base35/NimbusMonoPS-Italic.t1: Nimbus Mono PS:style=Italic /usr/share/fonts/urw-base35/NimbusMonoPS-Regular.t1: Nimbus Mono PS:style=Regular /usr/share/fonts/urw-base35/NimbusMonoPS-BoldItalic.t1: Nimbus Mono PS:style=Bold Italic /usr/share/fonts/urw-base35/NimbusMonoPS-Regular.otf: Nimbus Mono PS:style=Regular /usr/share/fonts/urw-base35/NimbusMonoPS-Bold.t1: Nimbus Mono PS:style=Bold
% fc-list "Noto Sans" | wc -l 72
On 5/23/22 13:05, home user wrote:
notice 4 fonts called "Noto Sans". They are different from each other. Notice 4 fonts called "Noto Sans, Bold". They are different from each other. There are other examples. This is only my opinion, but they should have different names. Is the problem in the font tools or in the fonts themselves? If the problem is in the fonts themselves, then there should be a standard requiring unique naming of fonts.
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
Terrible GUI design if it doesn't give you some indication that the name's been truncated. Are users supposed to around clicking on everything, now?
On 5/23/22 22:45, Tim via users wrote:
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
Terrible GUI design if it doesn't give you some indication that the name's been truncated. Are users supposed to around clicking on everything, now?
I don't think it's truncated. I suspect it's something to do with the way the font metadata is defined, but I don't know for sure. There are other fonts where that information is shown, so it's not the font app that's hiding it.
On 5/24/22 12:31 AM, Samuel Sieb wrote:
On 5/23/22 22:45, Tim via users wrote:
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
That's how I knew they were 4 different fonts.
Terrible GUI design if it doesn't give you some indication that the name's been truncated. Are users supposed to around clicking on everything, now?
I don't think it's truncated. I suspect it's something to do with the way the font metadata is defined, but I don't know for sure. There are other fonts where that information is shown, so it's not the font app that's hiding it.
Either way, there is a problem here. Yet I suspect that a bug is not the way to get this addressed. This is a significant problem, and will probably take a serious task/project to fix it. ...a font clean-up task/project. How do I propose it?
On 5/24/22 12:41, home user wrote:
On 5/24/22 12:31 AM, Samuel Sieb wrote:
On 5/23/22 22:45, Tim via users wrote:
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
That's how I knew they were 4 different fonts.
Terrible GUI design if it doesn't give you some indication that the name's been truncated. Are users supposed to around clicking on everything, now?
I don't think it's truncated. I suspect it's something to do with the way the font metadata is defined, but I don't know for sure. There are other fonts where that information is shown, so it's not the font app that's hiding it.
Either way, there is a problem here. Yet I suspect that a bug is not the way to get this addressed. This is a significant problem, and will probably take a serious task/project to fix it. ...a font clean-up task/project. How do I propose it?
I don't see the problem, but if you really have a problem with how the Noto Sans fonts show up in the app, then either file a bug on the font or the app.
On 5/24/22 4:37 PM, Samuel Sieb wrote:
On 5/24/22 12:41, home user wrote:
On 5/24/22 12:31 AM, Samuel Sieb wrote:
On 5/23/22 22:45, Tim via users wrote:
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
[... snip ...]
I don't see the problem, but if you really have a problem with how the Noto Sans fonts show up in the app, then either file a bug on the font or the app.
If I understand George's fc-list idea correctly, I have... ----- bash.8[~]: fc-list | wc -l 571 bash.9[~]: ----- ...571 fonts on this work station. The fonts I used in this thread are randomly-chosen representative examples of the problems I'm experiencing in one of my projects. I'd be somewhat surprised if there were not other fonts with the same problems. Samuel's reasoning that the problems are in the fonts rather than the tools makes sense to me. So a bug against the font tools (I know of 2, there ought to be more) does not make sense. If there are other fonts with the same problems, then a bug against one does not make sense; it would miss the other fonts. I do not know how to check all the fonts, and I don't have the time to (that project of mine is already way behind and quite under-estimated). That is why I think something else is needed, why I said what I said, and why I asked what I asked. I was hoping one of the Fedora chiefs on this list (Matthew Miller? Kevin Fenzi?) would chime in.
On Tue, May 24, 2022 at 4:41 PM home user mattisonw@comcast.net wrote:
On 5/24/22 12:31 AM, Samuel Sieb wrote:
On 5/23/22 22:45, Tim via users wrote:
On Mon, 2022-05-23 at 21:23 -0700, Samuel Sieb wrote:
You're not seeing the whole name. If you click on them, you'll see they are different. Regular and 3 types of condensed.
That's how I knew they were 4 different fonts.
Terrible GUI design if it doesn't give you some indication that the name's been truncated. Are users supposed to around clicking on everything, now?
I don't think it's truncated. I suspect it's something to do with the way the font metadata is defined, but I don't know for sure. There are other fonts where that information is shown, so it's not the font app that's hiding it.
Either way, there is a problem here. Yet I suspect that a bug is not the way to get this addressed. This is a significant problem, and will probably take a serious task/project to fix it. ...a font clean-up task/project. How do I propose it?
Unfortunately, widely used open source fonts often get multiple forks -- some that add support for additional languages, some that "improve" details of shapes and metrics. You end up with a font that is needed by those who require a certain language but that also cause problems with some tools or when used in existing documents that don't embed fonts due to changes in metrics.
Fonts are essentially programs, and like programs, you can have multiple forks with conflicting goals and coding styles that rely on features outside the language specification.
If you use open-source fonts that attempt to replace legacy PostScript (Laserwriter 35) fonts you run risks if you try to use them for things that you couldn't do with the original fonts.
Font creators now often include license clauses to require name changes for modified versions.
I have a few minutes to come back to this.
It seems to me that the font tools do not have a minor bug. Rather, they were probably fine years ago when fonts were fewer and generally simpler. Now there are more fonts, and they seem to be more complicated. So either the font tools need major enhancement or new tools are needed.
I don't have the knowledge of fonts and fontology(?) to propose a good, nearly complete set of requirements. But a few things do come to mind, such a ability to search, sort, and filter based on: * characteristics such as serif vs. sans serif. vs. regular (like Chinese kaiti) vs. cursive vs. (etc.). * characteristics such as Type-1, Type-2, Open, True, etc. * font name (font family?). * character height and character width. * stroke characteristics. * inclusion of characters from specific languages. Surely readers of this can think of other things.
A bug does not make sense for this. I think we've exhausted what can be done with what is currently available. So I'm marking this "CLOSED", though people are free to comment further. I thank those who contributed for their time and efforts.
On Sat, Jun 4, 2022 at 3:33 PM home user mattisonw@comcast.net wrote:
I have a few minutes to come back to this.
It seems to me that the font tools do not have a minor bug. Rather, they were probably fine years ago when fonts were fewer and generally simpler. Now there are more fonts, and they seem to be more complicated. So either the font tools need major enhancement or new tools are needed.
I don't have the knowledge of fonts and fontology(?) to propose a good, nearly complete set of requirements. But a few things do come to mind, such a ability to search, sort, and filter based on:
- characteristics such as serif vs. sans serif. vs. regular (like
Chinese kaiti) vs. cursive vs. (etc.).
- characteristics such as Type-1, Type-2, Open, True, etc.
- font name (font family?).
- character height and character width.
- stroke characteristics.
- inclusion of characters from specific languages.
Surely readers of this can think of other things.
I used to use xfontsel which has a search mechanism as you described. But with fewer search options (for obvious reasons for the time) and probably obsoleted by the various new font usages (display managers and printers, etc.)
Yet I see that on my system
$ fc-list | wc -l 600 xlsfonts|wc -l 1088
so I don't really know how many fonts are on my system nor how many are visible by which tool.
A bug does not make sense for this. I think we've exhausted what can be
done with what is currently available. So I'm marking this "CLOSED", though people are free to comment further. I thank those who contributed for their time and efforts.
users@lists.stg.fedoraproject.org