The main reason I suggested using Linux Libertine as the 'preferred' header font instead of DejuVu Serif is that DejaVu Serif doesn't at all resemble Georgia (the proprietary font the designers wanted). It's almost a slab-serif font, which is very different from the book/antique serifs of fonts like Georgia and Times. The designers wanted a classic, "encyclopedic" looking font for the headers. DejaVu Serif is a workhorse font, not a presentation/display font. It's appeal lies in its support for thousands of Unicode characters, definitely not in its design (which is frankly pretty awful). On pretty much any Linux system, specifying DejaVu Serif in the CSS is going to be redundant anyway, since most of the popular Linux installs fall back to DejaVu Serif as the default 'serif' font anyway. Thus any characters that can't get rendered in the preferred fonts will get rendered in DejaVu anyway.
When choosing the order of fonts in a CSS stack, you always put the prettiest ones first, and the most widely installed ones last. In the current case, we're doing it backwards.
I'm also open to using 'Nimbus Roman No9 L', which is more widely installed than Linux Libertine, but a lot nicer to look at than DejaVu Serif. It's basically a Times Roman replacement. Another option would be Liberation Serif, which is pretty ugly, but still nicer than DejaVu Serif.
My preferred font-stack for the headers would be: font-family: "Linux Libertine", Georgia, "Nimbus Roman No9 L", serif;
Really though, the preferred font should be chosen by the designers, not by a committee of developers, IMO.
Ryan Kaldari
Ryan, what about the argument that the font system will automatically do a replacement for Helvetia neue, rather than going to the next font in the list, effectively putting the choice in some Linux developers hands rather that ours?
Sent while mobile
On Dec 20, 2013, at 5:51 PM, Ryan Kaldari rkaldari@wikimedia.org wrote:
The main reason I suggested using Linux Libertine as the 'preferred' header font instead of DejuVu Serif is that DejaVu Serif doesn't at all resemble Georgia (the proprietary font the designers wanted). It's almost a slab-serif font, which is very different from the book/antique serifs of fonts like Georgia and Times. The designers wanted a classic, "encyclopedic" looking font for the headers. DejaVu Serif is a workhorse font, not a presentation/display font. It's appeal lies in its support for thousands of Unicode characters, definitely not in its design (which is frankly pretty awful). On pretty much any Linux system, specifying DejaVu Serif in the CSS is going to be redundant anyway, since most of the popular Linux installs fall back to DejaVu Serif as the default 'serif' font anyway. Thus any characters that can't get rendered in the preferred fonts will get rendered in DejaVu anyway.
When choosing the order of fonts in a CSS stack, you always put the prettiest ones first, and the most widely installed ones last. In the current case, we're doing it backwards.
I'm also open to using 'Nimbus Roman No9 L', which is more widely installed than Linux Libertine, but a lot nicer to look at than DejaVu Serif. It's basically a Times Roman replacement. Another option would be Liberation Serif, which is pretty ugly, but still nicer than DejaVu Serif.
My preferred font-stack for the headers would be: font-family: "Linux Libertine", Georgia, "Nimbus Roman No9 L", serif;
Really though, the preferred font should be chosen by the designers, not by a committee of developers, IMO.
Ryan Kaldari _______________________________________________ Design mailing list Design@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/design
Hi,
It seems you’re debating substituting fonts that have little in common: geometric slab serif vs. transitional serif or rational serif, hinted vs. unhinted, wide language and phonetic coverage vs. European language coverage, open source and maintained vs. unmodifiable.
Maybe if you defined criteria, you'd be able to choose with empirical arguments rather than just throwing likes and don’t-likes.
Have you considered using CSS3 @font-face webfonts? Using webfonts could mean the same font is used on all platforms with similar outcome (style, quality, coverage, maintainability).
Cheers,
On Sat, Dec 21, 2013 at 9:30 AM, Jared Zimmerman jzimmerman@wikimedia.org wrote:
Ryan, what about the argument that the font system will automatically do a replacement for Helvetia neue, rather than going to the next font in the list, effectively putting the choice in some Linux developers hands rather that ours?
Sent while mobile
On Dec 20, 2013, at 5:51 PM, Ryan Kaldari rkaldari@wikimedia.org wrote:
The main reason I suggested using Linux Libertine as the 'preferred' header font instead of DejuVu Serif is that DejaVu Serif doesn't at all resemble Georgia (the proprietary font the designers wanted). It's almost a slab-serif font, which is very different from the book/antique serifs of fonts like Georgia and Times. The designers wanted a classic, "encyclopedic" looking font for the headers. DejaVu Serif is a workhorse font, not a presentation/display font. It's appeal lies in its support for thousands of Unicode characters, definitely not in its design (which is frankly pretty awful). On pretty much any Linux system, specifying DejaVu Serif in the CSS is going to be redundant anyway, since most of the popular Linux installs fall back to DejaVu Serif as the default 'serif' font anyway. Thus any characters that can't get rendered in the preferred fonts will get rendered in DejaVu anyway.
When choosing the order of fonts in a CSS stack, you always put the prettiest ones first, and the most widely installed ones last. In the current case, we're doing it backwards.
I'm also open to using 'Nimbus Roman No9 L', which is more widely installed than Linux Libertine, but a lot nicer to look at than DejaVu Serif. It's basically a Times Roman replacement. Another option would be Liberation Serif, which is pretty ugly, but still nicer than DejaVu Serif.
My preferred font-stack for the headers would be: font-family: "Linux Libertine", Georgia, "Nimbus Roman No9 L", serif;
Really though, the preferred font should be chosen by the designers, not by a committee of developers, IMO.
Ryan Kaldari _______________________________________________ Design mailing list Design@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/design
Design mailing list Design@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/design
On Sat, Dec 21, 2013 at 4:09 AM, Denis Jacquerye moyogo@gmail.com wrote:
Hi,
It seems you’re debating substituting fonts that have little in common: geometric slab serif vs. transitional serif or rational serif, hinted vs. unhinted, wide language and phonetic coverage vs. European language coverage, open source and maintained vs. unmodifiable.
Maybe if you defined criteria, you'd be able to choose with empirical arguments rather than just throwing likes and don’t-likes.
Have you considered using CSS3 @font-face webfonts? Using webfonts could mean the same font is used on all platforms with similar outcome (style, quality, coverage, maintainability).
Yes, the problem is that we have two mutually exclusive requirements and no communication between the two opposing camps. On one side, the designers have said that the preferred body font should be Helvetica Neue (or Helvetica) and the header font should be Georgia. On the other side we have some vocal community members and developers who say that the preferred font must be a free font (but they don't say which). So what we end up with is a totally inconsistent experience which defeats the entire purpose of the typography update.
I agree that what we really need is a list of font requirements, not a list of fonts. This list should include both aesthetic and technical requirements. For example:
*Header Font Requirements:* * Includes all glyphs for Basic Latin (ASCII) and Latin-1 Supplement (0080–00FF) * Includes at least 90% of Latin Extended-A (0100–017F) * Is installed by default on at least 1 common operating system * Easy to read at large and medium text sizes * Non-distracting design (readers should not notice the font) * Traditional serifs (for an "authoritative", encyclopedic look)
*Body Font Requirements:* * Includes all glyphs for Basic Latin (ASCII), Latin-1 Supplement (0080–00FF), and Latin Extended-A (0100–017F) * Includes at least 90% of Latin Extended-B (0180–024F) and IPA Extensions (0250–02AF) * Is installed by default on at least 1 common operating system * Renders common ligatures correctly (many fonts do not) * Easy to read at medium, small, and very small text sizes (tall x-height) * Non-distracting design (readers should not notice the font) * Sans-serif * Uses most common variants of variable Latin letters, for example, two story lower-case "a" and single-story lower-case "g"
These are just some examples of course. If we had such a list of requirements, we could hopefully use it to construct a font-stack that is consistent and that everyone (or at least more than one person) agrees on :)
Ryan Kaldari
On 23 Dec 2013 17:18, "Ryan Kaldari" rkaldari@wikimedia.org wrote:
On the other side we have some vocal community members and developers who
say that the preferred font must be a free font (but they don't say which). So what we end up with is a totally inconsistent experience which defeats the entire purpose of the typography update.
This. We should go back to the community asking for this and get them to decide on a suitable free font similar to Helvetica since they have brought us into this discussion. Right now it seems like the camp that is agnostic about free fonts (myself included) is struggling to find a suitable alternative which seems unfair. If someone wants to see something happen they should be prepared to help make that happen.
On 12/23/2013 06:06 PM, Jon Robson wrote:
On 23 Dec 2013 17:18, "Ryan Kaldari" rkaldari@wikimedia.org wrote:
On the other side we have some vocal community members and developers who
say that the preferred font must be a free font (but they don't say which). So what we end up with is a totally inconsistent experience which defeats the entire purpose of the typography update.
This. We should go back to the community asking for this and get them to decide on a suitable free font similar to Helvetica since they have brought us into this discussion. Right now it seems like the camp that is agnostic about free fonts (myself included) is struggling to find a suitable alternative which seems unfair. If someone wants to see something happen they should be prepared to help make that happen.
To be precise, what brought us into this discussion was the idea of committing to certain proprietary fonts.
The vocal community members are saying that Wikimedia / MediaWiki shouldn't be pointing to any proprietary product when there are free alternatives available. We just ask the designers to stick to the same principles of free knowledge and free software everybody else is following.
This is why this "side" has no strong opinion on specific fonts. We agree that defining fonts is a task for designers, but still a task to be performed under the same Wikimedia principles.
The recurrent argument from the pro-proprietary-fonts side is that no free font is as good and has the wide install base. Following this argument we would have never started a free encyclopedia from scratch, because in 2001 anybody with a computer would most likely have access to better proprietary encyclopedias with wide distribution.
But no, someone had a vision of a free encyclopedia that anyone can edit, and now we have an amazing product. Following the same principle and the same bet we should go for free fonts that anyone can edit. Fonts that luckily we can choose (and eventually improve) without needing to create them from scratch. We should do this despite the short term inconveniences, aiming for the long term goal of a World with amazing free fonts for everybody, perfectly suitable for our amazing projects.
This is the first part of the discussion. If we don't agree on the points above then we will probably keep arguing about the points below.
If we agree on the points above then we have basically two strong choices to reach consensus:
1. Choose the free fonts we want to commit to, seeing these fonts not only as design objects but also as living open source projects we could promote and get involved with. We should considering the translations these free fonts have in systems missing them, but not being determined by them.
2. Don't commit to any font. Point to serif / sans and let the browsers do the rest.
According to the discussion so far, there is a possible third choice (a weak one) to find a compromise, choosing some free and non-free fonts. I consider this argument weak because I believe it is originated by a lack of consensus in the points above about the free knowledge mission and its relation with font choices.
PS: if you really need a starting point for the selection of free fonts, two typefaces that the Wikimedia community has already selected in the past are Linux Libertine and Gill Sans -- https://wikimediafoundation.org/wiki/Visual_identity_guidelines
Just one thing: I expressed the idea of sticking to free fonts only because I *personally* believe it is the right thing to do. I have followed the discussion, I have tested the beta, this is my opinion, and I just wanted to expose it.
That's it. I'm no stakeholder in this task and I respect enormously the effort that many great people is putting in solving this puzzle of priorities. If the point of the "free fonts only" side is taken then I'm happy enough with whatever it is decided.
The motivation of my reply was simply to clarify the bottom line of the free-font reasoning. I hope it was useful. If not, please ignore.
Thank you for your patience and your work.
On 12/24/2013 01:28 PM, Quim Gil wrote:
On 12/23/2013 06:06 PM, Jon Robson wrote:
On 23 Dec 2013 17:18, "Ryan Kaldari" rkaldari@wikimedia.org wrote:
On the other side we have some vocal community members and developers who
say that the preferred font must be a free font (but they don't say which). So what we end up with is a totally inconsistent experience which defeats the entire purpose of the typography update.
This. We should go back to the community asking for this and get them to decide on a suitable free font similar to Helvetica since they have brought us into this discussion. Right now it seems like the camp that is agnostic about free fonts (myself included) is struggling to find a suitable alternative which seems unfair. If someone wants to see something happen they should be prepared to help make that happen.
To be precise, what brought us into this discussion was the idea of committing to certain proprietary fonts.
The vocal community members are saying that Wikimedia / MediaWiki shouldn't be pointing to any proprietary product when there are free alternatives available. We just ask the designers to stick to the same principles of free knowledge and free software everybody else is following.
This is why this "side" has no strong opinion on specific fonts. We agree that defining fonts is a task for designers, but still a task to be performed under the same Wikimedia principles.
The recurrent argument from the pro-proprietary-fonts side is that no free font is as good and has the wide install base. Following this argument we would have never started a free encyclopedia from scratch, because in 2001 anybody with a computer would most likely have access to better proprietary encyclopedias with wide distribution.
But no, someone had a vision of a free encyclopedia that anyone can edit, and now we have an amazing product. Following the same principle and the same bet we should go for free fonts that anyone can edit. Fonts that luckily we can choose (and eventually improve) without needing to create them from scratch. We should do this despite the short term inconveniences, aiming for the long term goal of a World with amazing free fonts for everybody, perfectly suitable for our amazing projects.
This is the first part of the discussion. If we don't agree on the points above then we will probably keep arguing about the points below.
If we agree on the points above then we have basically two strong choices to reach consensus:
- Choose the free fonts we want to commit to, seeing these fonts not
only as design objects but also as living open source projects we could promote and get involved with. We should considering the translations these free fonts have in systems missing them, but not being determined by them.
- Don't commit to any font. Point to serif / sans and let the browsers
do the rest.
According to the discussion so far, there is a possible third choice (a weak one) to find a compromise, choosing some free and non-free fonts. I consider this argument weak because I believe it is originated by a lack of consensus in the points above about the free knowledge mission and its relation with font choices.
PS: if you really need a starting point for the selection of free fonts, two typefaces that the Wikimedia community has already selected in the past are Linux Libertine and Gill Sans -- https://wikimediafoundation.org/wiki/Visual_identity_guidelines
On 25/12/13 06:59, Quim Gil wrote:
Just one thing: I expressed the idea of sticking to free fonts only because I *personally* believe it is the right thing to do. I have followed the discussion, I have tested the beta, this is my opinion, and I just wanted to expose it.
That's it. I'm no stakeholder in this task and I respect enormously the effort that many great people is putting in solving this puzzle of priorities. If the point of the "free fonts only" side is taken then I'm happy enough with whatever it is decided.
The motivation of my reply was simply to clarify the bottom line of the free-font reasoning. I hope it was useful. If not, please ignore.
I hope it will be considered useful. Those were good points to make.
On Sat, Dec 21, 2013 at 1:30 AM, Jared Zimmerman jzimmerman@wikimedia.orgwrote:
Ryan, what about the argument that the font system will automatically do a replacement for Helvetia neue
You mean "Helvetica"; but we're talking about the Georgia, serif font font stack for headers.
, rather than going to the next font in the list, effectively putting the choice in some Linux developers hands rather that ours?
On Ubuntu, the equivalent font substitutions are Nimbus Roman No9 L for "Times", and Tinos, Liberation Serif, Thorndale, and Thorndale AMT for "Times New Roman"; of the latter set I only have Liberation Serif (note it's not the Linux Libertine that Kaldari recommends), so I get that one. So the equivalent of
font-family: "Helvetica Neue", "Helvetica", "Nimbus Sans L", "Arial", "Liberation Sans", sans-serif;
is something like:
font-family: "Georgia", "Times", "Linux Libertine", serif;
As I said when we discussed sans-serif font stack ([Wikitech-l] Should MediaWiki CSS prefer non-free fonts?), if we prefer the open source font (for the criteria Denis Jacquerye mentions), then we should list it before the proprietary font. The problem remains: getting designers and l10n advisers to install and verifiably use these open source fonts on Mac and Windows in order to make the call.
Nobody responded to http://jsfiddle.net/UPBUH/7/ with useful info on what Android, Mac, or Windows does, but it's still there if someone wants to add the serif fonts.