830002 - text-transform: uppercase isn't good

Status: RESOLVED WONTFIX

(Firefox OS Graveyard :: Gaia, defect)

Description

[Axel Hecht [:Pike]]

Using text-transform: uppercase makes a decision on UX that's spanning locales. We shouldn't do that.

I also suspect that this is what's making bug 828311 looking unfixed right now.

Willyaranda, is this a blocker-issue for Spanish?

Reuben, Fabio, how about pt-BR?

Comment 1

[Reuben Morais [:reuben]]

If you're wondering, here are some places where |text-transform: uppercase| is used:

Calendar: Names of days
Clock: "Hours" and "minutes" in the new alarm notification.
Contacts: Names of numbers and addresses (home/personal/work/etc)
SMS: Dates

(In reply to Axel Hecht [:Pike] from comment #0)
> Reuben, Fabio, how about pt-BR?

No, this is not a blocker for pt-BR.

Comment 2

[Guillermo López :willyaranda (probably SLOW response)]

(In reply to Reuben Morais [:reuben] from comment #1)
> If you're wondering, here are some places where |text-transform: uppercase|
> is used:
> 
> Calendar: Names of days
> Clock: "Hours" and "minutes" in the new alarm notification.
> Contacts: Names of numbers and addresses (home/personal/work/etc)
> SMS: Dates
> 
> (In reply to Axel Hecht [:Pike] from comment #0)
> > Reuben, Fabio, how about pt-BR?
> 
> No, this is not a blocker for pt-BR.

Neither for Spanish.

Comment 3

[Staś Małolepszy :stas]

Looks like this is okay for 1.0, but we should fix it for 1.1.  text-transform: uppercase will break Greek, which is one of our target 1.1 locales.  See bug 667430 for reference.

Comment 4

[Jaison Justus]

I am a newbee to Gaia, can i tryout this bug.

Comment 5

[Francesco Lodolo [:flod]]

I currently count 38 instances of "text-transform: uppercase" across apps. Some of them are in test apps, but some others are in building blocks.(In reply to Staś Małolepszy :stas from comment #3)

> See bug 667430 for reference.
I'm not sure that's a good example: we *can't* use text-transform on websites because we don't have control over font and browser support. But for Firefox OS we control both the font and the rendering engine, so that's not an issue.

Having said that, I still agree with comment 0.

Comment 6

[Francesco Lodolo [:flod]]

(In reply to Francesco Lodolo [:flod] from comment #5)
> I'm not sure that's a good example: we *can't* use text-transform on
> websites because we don't have control over font and browser support. But
> for Firefox OS we control both the font and the rendering engine, so that's
> not an issue.

Correcting myself after reading tonight's bugmail
https://bugzilla.mozilla.org/show_bug.cgi?id=854756#c10

Comment 7

[Julien Wajsberg [:julienw]]

bug 854756 comment 10 is on the Web, right? It's not an app.

If there is an issue with text-transform, can't we fix it in the platform?

Comment 8

[Francesco Lodolo [:flod]]

(In reply to Julien Wajsberg [:julienw] from comment #7)
> bug 854756 comment 10 is on the Web, right? It's not an app.

It's on mozilla.org, but it means we'd have the same issue on Firefox OS.

Kevin, have you already seen issues on Firefox OS related to text-transform?

Comment 9

[Julien Wajsberg [:julienw]]

Yes, I mean that it's normal to make this work properly on all browsers for mozilla.org, but in Firefox OS, if we can fix the platform to make this work properly, it's way better.

Comment 10

[Panos Astithas (he/him) [:past] (please ni?)]

(In reply to Staś Małolepszy :stas from comment #3)
> Looks like this is okay for 1.0, but we should fix it for 1.1. 
> text-transform: uppercase will break Greek, which is one of our target 1.1
> locales.  See bug 667430 for reference.

We fixed "text-transform: uppercase" and "font-variant: small-caps" in bug 307039 which predates Firefox OS 1.0. If the "lang" attribute is properly set on the DOM element per spec Gecko will do the right thing.

Comment 11

[Francesco Lodolo [:flod]]

I'm wondering about this comment
https://bugzilla.mozilla.org/show_bug.cgi?id=667430#c4

Comment 12

[Panos Astithas (he/him) [:past] (please ni?)]

That comment predates the fix by about a year ;-)

Comment 13

[Francesco Lodolo [:flod]]

(In reply to Panos Astithas [:past] from comment #12)
> That comment predates the fix by about a year ;-)

Argh, I should stop looking just at bug numbers.

CCing also Jonathan who may have a better idea of our current status (also curious about Indic locales).

Comment 14

[Kevin Scannell]

(In reply to Francesco Lodolo [:flod] from comment #8)
> Kevin, have you already seen issues on Firefox OS related to text-transform?

No, but only through good luck... there are lots of translations that would be subject to this issue were "text-transform: uppercase" to be applied; you can grep for them with this:

$ find . -name '*.properties' | xargs egrep '=[^{]+[bdghmnt][A-Z]'

Comment 15

[Jonathan Kew [:jfkthame]]

We fixed things for Greek, AFAIK, but using text-transform:uppercase for arbitrary Irish content is still a problem. It's apparently common to keep certain initial (prefix) letters in lowercase even when the text is otherwise all-caps; see for example bug 854756 comment 10, and http://en.wikipedia.org/wiki/Irish_orthography#Capitalisation. (It's not clear to me whether this is an absolute rule or a stylistic preference; the wording in the Wikipedia article seems to imply the latter.)

Kevin, is there any way such prefixes could reliably be identified by a capitalization process, so as to treat them appropriately? (I'm guessing not, but I don't know anything about the language.)

If not, then I don't see any way this can be "fixed" in the platform; it can only be fixed by cooperation between authors/designers and localizers.

Note that text-transform can be problematic for other languages, too. Even in English, there are certain examples that might merit care. While "MacMillan" can reasonably be rendered as "MACMILLAN" in all-caps, the spelling "McMillan" doesn't work so well: "MCMILLAN" is difficult to read. Might be better as "McMILLAN", or "MᴄMILLAN" with a small-cap ᴄ (your rendering may vary, depending on fonts!) I don't think the platform can be expected to magically get things like this "right", however; authors and designers need to take responsibility for such content-dependent refinements.

(BTW, Francesco: this shouldn't be an issue for Indic locales, as the Indic scripts don't have case.)

Comment 16

[Francesco Lodolo [:flod]]

(In reply to Jonathan Kew (:jfkthame) from comment #15)
> (BTW, Francesco: this shouldn't be an issue for Indic locales, as the Indic
> scripts don't have case.)

Thanks, I never fully realized that for Indic locales (and, on second thoughts, I should have).

Comment 17

[Kevin Scannell]

(In reply to Jonathan Kew (:jfkthame) from comment #15)
> We fixed things for Greek, AFAIK, but using text-transform:uppercase for
> arbitrary Irish content is still a problem. It's apparently common to keep
> certain initial (prefix) letters in lowercase even when the text is
> otherwise all-caps; see for example bug 854756 comment 10, and
> http://en.wikipedia.org/wiki/Irish_orthography#Capitalisation. (It's not
> clear to me whether this is an absolute rule or a stylistic preference; the
> wording in the Wikipedia article seems to imply the latter.)

There is a published "Official Standard" for Irish, here:

http://www.oireachtas.ie/parliament/media/Final-Version.pdf

Interestingly, while the standard doesn't have an explicit rule for typesetting in allcaps, the section headings in the book itself are allcaps and they use lowercase letters in the cases we're discussing!  That's about as close to a rule as we're going to get :)   

This is done in part to avoid ambiguities, i.e. not purely for style or readability; for example:

Athair = Father
Nathair = Snake
Ár nAthair = Our Father
Ár Nathair = Our Snake
ÁR NATHAIR = OUR SNAKE
ÁR nATHAIR = OUR FATHER

One does see a stylistic choice between lowercase and smallcaps for the prefix letter(s); both are acceptable to me.   

> 
> Kevin, is there any way such prefixes could reliably be identified by a
> capitalization process, so as to treat them appropriately? (I'm guessing
> not, but I don't know anything about the language.)

Yes. It's (almost) completely algorithmic.  There's one edge case with a prefixed h but 99%+ of cases are handled with a list of ~20 patterns that I can provide easily (shall I file a new bug?).

Comment 18

[Jonathan Kew [:jfkthame]]

(In reply to Kevin Scannell from comment #17)
> > Kevin, is there any way such prefixes could reliably be identified by a
> > capitalization process, so as to treat them appropriately? (I'm guessing
> > not, but I don't know anything about the language.)
> 
> Yes. It's (almost) completely algorithmic.  There's one edge case with a
> prefixed h but 99%+ of cases are handled with a list of ~20 patterns that I
> can provide easily (shall I file a new bug?).

Please do. Obviously, I can't promise a quick fix, but having the details on file is the first step. Perhaps we can do something similar to what we did for Greek, once we understand the requirements more clearly.

Comment 19

[:Gijs (he/him)]

(In reply to Jonathan Kew (:jfkthame) from comment #18)
> (In reply to Kevin Scannell from comment #17)
> > > Kevin, is there any way such prefixes could reliably be identified by a
> > > capitalization process, so as to treat them appropriately? (I'm guessing
> > > not, but I don't know anything about the language.)
> > 
> > Yes. It's (almost) completely algorithmic.  There's one edge case with a
> > prefixed h but 99%+ of cases are handled with a list of ~20 patterns that I
> > can provide easily (shall I file a new bug?).
> 
> Please do. Obviously, I can't promise a quick fix, but having the details on
> file is the first step. Perhaps we can do something similar to what we did
> for Greek, once we understand the requirements more clearly.

Did this happen?

Comment 20

[Jonathan Kew [:jfkthame]]

For Irish, we fixed text-transform:uppercase in bug 1014639.

Regarding the appropriateness of text-transform styling in general, I'm sure there are other languages/locales where similar issues apply. So authors still need to be cautious about applying such properties, especially where a design is intended to work across multiple, perhaps unknown languages.