[css-text] I18N-ISSUE-331: No kashida style or relationship to styles 'distribute' and 'inter-word'

State: OPEN WG Comment Product: CSS3-text Raised by: Addison Phillips Opened on: 2014-01-23 Description: Section 7.3: Surprisingly there is no keyword specifically associated with kashida (other than "auto"). Shouldn't there be? Or is it a special case of "distribute"? Some WG members [1] thought that "inter-word" implies turning kashida off. [1] http://www.w3.org/2013/12/05-i18n-minutes.html

There was a request in Tucson F2F[1] that: * “auto” automagically does the best justification for the content language * Define values only when multiple choices of justification behavior are required within one language * That behavior should be described so that implementers who do not know the language can implement in an interoperable way and with our knowledge, the “kashida” value did not meet the criteria. The discussion started to cut values down to “auto | distribute”[2], then we had information that “none” and “inter-word” meet the criteria. Can you give such information for “kashida”? [1] http://lists.w3.org/Archives/Public/www-style/2013Feb/0474.html [2] http://lists.w3.org/Archives/Public/www-style/2012Dec/0057.html /koji On Jan 25, 2014, at 3:30 AM, Phillips, Addison <addison@lab126.com> wrote: > State: > OPEN WG Comment > Product: > CSS3-text > Raised by: > Addison Phillips > Opened on: > 2014-01-23 > Description: > Section 7.3: Surprisingly there is no keyword specifically associated with kashida (other than "auto"). Shouldn't there be? Or is it a special case of "distribute"? Some WG members [1] thought that "inter-word" implies turning kashida off. > > [1] http://www.w3.org/2013/12/05-i18n-minutes.html

Koji Ishii scripsit: > * “auto” automagically does the best justification for the content language Is there any room for disagreement about "the best"? The great bulk of printed English text is justified, but many people prefer ragged right. -- John Cowan http://www.ccil.org/~cowan cowan@ccil.org He that would foil me must use such weapons as I do, for I have not fed my readers with straw, neither will I be confuted with stubble. --Thomas Vaughan (1650)

Some comments follow. > > There was a request in Tucson F2F[1] that: > * “auto” automagically does the best justification for the content language Which is fine. Note that our comment is that *only* the "auto" justification selects "kashida" (if any do). > * Define values only when multiple choices of justification behavior are > required within one language Which is also "okay", insofar as it goes. But some of these justification forms are ill-suited to certain scripts or languages. And kashida certainly meets this criterion: several languages might need to select it. > * That behavior should be described so that implementers who do not know the > language can implement in an interoperable way and with our knowledge, the > “kashida” value did not meet the criteria. Kashida is not currently well-described in English, so far as I know. I have it on good authority that there is at least one book in Persian on the topic. I am unaware (having neither seen the book nor the ability to read Persian) of whether this would apply generally or only to that language family, or how well this document succeeds at its task. So I agree that kashida does not meet this criterion presently. But, since you mention it in the text (you even have an example of Tasmeem rendering it), it seems like an oversight to mention it and even encourage it in "auto" but not do say anything further (such as whether it is on or off in one of the other modes or if those modes). > > The discussion started to cut values down to “auto | distribute”[2], then we had > information that “none” and “inter-word” meet the criteria. I agree that these all meet the criteria you established. The question we asked is what the effect of some of the other keywords (notably inter-word) would be on a kashida implementation (as well as whether a kashida type or types could be created later). > > Can you give such information for “kashida”? See above. We would like to, but are not in a position to. Addison

On 2014/04/21 04:11, John Cowan wrote: > Koji Ishii scripsit: > >> * “auto” automagically does the best justification for the content language > > Is there any room for disagreement about "the best"? The great bulk of > printed English text is justified, but many people prefer ragged right. Yes. And it depends on the situation. On the Web, ragged right is so popular as to be close to "the best", even if that's just a result of (former?) browser technology limitations. Regards, Martin.

> > > >> * “auto” automagically does the best justification for the content > >> language > > > > Is there any room for disagreement about "the best"? The great bulk > > of printed English text is justified, but many people prefer ragged right. > > Yes. And it depends on the situation. On the Web, ragged right is so popular as > to be close to "the best", even if that's just a result of > (former?) browser technology limitations. > Sure, but presumably "auto" would not be a synonym for "none"...

>>>> * “auto” automagically does the best justification for the content >>>> language >>> >>> Is there any room for disagreement about "the best"? The great bulk >>> of printed English text is justified, but many people prefer ragged right. >> >> Yes. And it depends on the situation. On the Web, ragged right is so popular as >> to be close to "the best", even if that's just a result of >> (former?) browser technology limitations. >> > > Sure, but presumably "auto" would not be a synonym for "none"... That’s not a problem here because this discussion is about text-justify:auto, which has no effect unless the author sets text-align:justify. Authors can choose ragged right by text-align:start. To make this visual tree: text-align:start —> ragged right text-align:justify text-justify:auto —> automagically does the best justification for the content language text-justify:<other values> —> specific justification behavior /koji

On Apr 21, 2014, at 6:59 AM, Phillips, Addison <addison@lab126.com> wrote: >> There was a request in Tucson F2F[1] that: >> * “auto” automagically does the best justification for the content language > > Which is fine. Note that our comment is that *only* the "auto" justification selects "kashida" (if any do). So, is the suggestion to recommend the use of kashida for “auto” if the content language is Arabic? I’m sorry for my ignorance but I don’t have a good sense whether kashida should be on by default or not, for what content languages. If I18N could recommend anything, I could try to put it in as an informative text, just like we refer to JLREQ for Japanese. Note that we have the following informative text already[1]: > using cursive elongation for Arabic Isn’t this kashida? >> * Define values only when multiple choices of justification behavior are >> required within one language > > Which is also "okay", insofar as it goes. But some of these justification forms are ill-suited to certain scripts or languages. And kashida certainly meets this criterion: several languages might need to select it. Yeah, that’s my understanding too. MS Word has 3 options for kashida[2], but I’m unable to justify them, nor specify the behavior. It looks like harfbuzz is willing to implement kashida someday[3], at that point we may be able to get better information on what options we should put into CSS. >> * That behavior should be described so that implementers who do not know the >> language can implement in an interoperable way and with our knowledge, the >> “kashida” value did not meet the criteria. > > Kashida is not currently well-described in English, so far as I know. I have it on good authority that there is at least one book in Persian on the topic. I am unaware (having neither seen the book nor the ability to read Persian) of whether this would apply generally or only to that language family, or how well this document succeeds at its task. > > So I agree that kashida does not meet this criterion presently. But, since you mention it in the text (you even have an example of Tasmeem rendering it), it seems like an oversight to mention it and even encourage it in "auto" but not do say anything further (such as whether it is on or off in one of the other modes or if those modes). Because I’m technically unable to…does I18N WG recommend to turn kashida off for “none" and "inter-word”, but on for “auto” and “distribute"? >> The discussion started to cut values down to “auto | distribute”[2], then we had >> information that “none” and “inter-word” meet the criteria. > > I agree that these all meet the criteria you established. The question we asked is what the effect of some of the other keywords (notably inter-word) would be on a kashida implementation (as well as whether a kashida type or types could be created later). >> >> Can you give such information for “kashida”? > > See above. We would like to, but are not in a position to. Sadly, me too, but you seem to have better knowledge than me. Your support is appreciated. [1] http://dev.w3.org/csswg/css-text/#text-justify-property [2] http://office.microsoft.com/en-gb/word-help/specify-kashida-length-in-arabic-text-HP005258100.aspx [3] https://groups.google.com/forum/#!topic/persian-computing/s-ftgmBvlF0 /koji

This is the most useful reference in English on Arabic text justification that I have come across: http://www.tug.org/TUGboat/tb27-2/tb87benatia.pdf James On Mon, Apr 21, 2014 at 4:59 AM, Phillips, Addison <addison@lab126.com>wrote: > Some comments follow. > > > > > There was a request in Tucson F2F[1] that: > > * “auto” automagically does the best justification for the content > language > > Which is fine. Note that our comment is that *only* the "auto" > justification selects "kashida" (if any do). > > > * Define values only when multiple choices of justification behavior are > > required within one language > > Which is also "okay", insofar as it goes. But some of these justification > forms are ill-suited to certain scripts or languages. And kashida certainly > meets this criterion: several languages might need to select it. > > > * That behavior should be described so that implementers who do not know > the > > language can implement in an interoperable way and with our knowledge, > the > > “kashida” value did not meet the criteria. > > Kashida is not currently well-described in English, so far as I know. I > have it on good authority that there is at least one book in Persian on the > topic. I am unaware (having neither seen the book nor the ability to read > Persian) of whether this would apply generally or only to that language > family, or how well this document succeeds at its task. > > So I agree that kashida does not meet this criterion presently. But, since > you mention it in the text (you even have an example of Tasmeem rendering > it), it seems like an oversight to mention it and even encourage it in > "auto" but not do say anything further (such as whether it is on or off in > one of the other modes or if those modes). > > > > The discussion started to cut values down to “auto | distribute”[2], > then we had > > information that “none” and “inter-word” meet the criteria. > > I agree that these all meet the criteria you established. The question we > asked is what the effect of some of the other keywords (notably inter-word) > would be on a kashida implementation (as well as whether a kashida type or > types could be created later). > > > > Can you give such information for “kashida”? > > See above. We would like to, but are not in a position to. > > Addison >

On 23/04/2014 13:55, James Clark wrote: > This is the most useful reference in English on Arabic text > justification that I have come across: > > http://www.tug.org/TUGboat/tb27-2/tb87benatia.pdf Fwiw, I also used to rely on that document for an understanding of the rules of justification in Arabic, however I've been chatting with Tom Milo (who developed Tasmeem and high end fonts and rendering for Arabic, and produces Koranic texts based on analysis of original documents), and he has some somewhat divergeant views on how justification should be achieved. I hope that we can get more involvement from Arabic experts to obtain some consensus on the situation (I proposed the idea of an Arabic Layout Requirements document on a trip to Oman this month). I suspect that it will however take some time to reach that consensus, and I also suspect that our discussion will unfortunately need to wait until then for resolution. RI

On 01/24/2014 10:30 AM, Phillips, Addison wrote: > State: > OPEN WG Comment > Product: > CSS3-text > Raised by: > Addison Phillips > Opened on: > 2014-01-23 > Description: > Section 7.3: Surprisingly there is no keyword specifically associated > with kashida (other than "auto"). Shouldn't there be? There is no special keyword for it, just as there is no special keyword for standard Japanese justification. > Or is it a special case of "distribute"? No. It is most definitely not a special case of "distribute". > Some WG members [1] thought that "inter-word" implies turning kashida off. This is true. It turns it off, because it only allows expansion at spaces. ~fantasai

[forwarding to www-international so that our tracker picks it up, apologies for duplication for those on www-style please reply to this email to keep the thread intact on both lists] -------- Original Message -------- Subject: Re: [css-text] I18N-ISSUE-331: No kashida style or relationship to styles 'distribute' and 'inter-word' Resent-Date: Fri, 09 May 2014 00:30:14 +0000 Resent-From: www-style@w3.org Date: Thu, 08 May 2014 17:29:44 -0700 From: fantasai <fantasai.lists@inkedblade.net> To: www-style@w3.org On 01/24/2014 10:30 AM, Phillips, Addison wrote: > State: > OPEN WG Comment > Product: > CSS3-text > Raised by: > Addison Phillips > Opened on: > 2014-01-23 > Description: > Section 7.3: Surprisingly there is no keyword specifically associated > with kashida (other than "auto"). Shouldn't there be? There is no special keyword for it, just as there is no special keyword for standard Japanese justification. > Or is it a special case of "distribute"? No. It is most definitely not a special case of "distribute". > Some WG members [1] thought that "inter-word" implies turning kashida off. This is true. It turns it off, because it only allows expansion at spaces. ~fantasai

I suspect we should say as little as possible for now in the spec about text elongation in Arabic script for justification of running text in paragraphs, since I believe that in the end the requirements will be somewhat complex, and we are a long way from understanding them yet. For example, it appears that some font types (eg. Ruqah style fonts) should not be used with elongation at all - all the justification is therefore done by manipulating space. Also, it's looking to me as if justification in, say, a naskh type font is likely to involve adjusting spacing a lot of the time, but may occasionally involve that plus some elongation when the spaces get too wide (eg. when a url is embedded in a narrow column). In other cases, however, a greater amount of elongation may be favoured in a naskh type font. Nastaliq font types, on the other hand, seem to rely more on elongation (and ligation and other shortening techniques) than naskh (since spaces are seen in a somewhat different light for nastaliq script). Also, btw, the rules for elongation are somewhat different in nastaliq than for naskh. Like detection of conjunct behaviour in indic scripts, the determination of the correct behaviour seems to be font dependent, and it's not clear to me how to handle that. Note that all this therefore also raises questions about how to do the right thing when a user changes the font by design or a font is substituted due to availability of fonts on a particular device). It suggests that maybe we need special controls for elongation that say something like: apply elongation if the font is X or Y only. (By the way, in passing, note that in Tasmeem it is possible to independently set parameters for the size of spaces inside a word (ie. where a letter doesn't join to the left) as well as outside a word.) I guess we can say that inter-word turns off elongation, since that is the purpose of that value. I'm not sure we can say much yet, until we understand things better, of the relationship between distribute and elongation, since elongation seems to involve far too many, poorly understood, complications. The current text says "Justification adjusts spacing between each pair of adjacent visually-perceived characters", which may be sufficient to deal with elongations, since joined cursive letters don't have any spaces between them. I suppose it introduces a question about stretching of gaps inside a word where letters don't join to the left, though. PS: For a discussion of nastaliq justification in Urdu see http://www.cle.org.pk/Publication/papers/2004/rule-based-expert-system.pdf > Resent-Date: Fri, 09 May 2014 00:30:14 +0000 > Resent-From: www-style@w3.org > Date: Thu, 08 May 2014 17:29:44 -0700 > From: fantasai <fantasai.lists@inkedblade.net> > To: www-style@w3.org > > On 01/24/2014 10:30 AM, Phillips, Addison wrote: >> State: >> OPEN WG Comment >> Product: >> CSS3-text >> Raised by: >> Addison Phillips >> Opened on: >> 2014-01-23 >> Description: >> Section 7.3: Surprisingly there is no keyword specifically >> associated >> with kashida (other than "auto"). Shouldn't there be? > > There is no special keyword for it, just as there is no special keyword > for standard Japanese justification. > >> Or is it a special case of "distribute"? > > No. It is most definitely not a special case of "distribute". > >> Some WG members [1] thought that "inter-word" implies turning >> kashida off. > > This is true. It turns it off, because it only allows expansion > at spaces. > > ~fantasai > > > >

> I suspect we should say as little as possible for now in the spec about text elongation in Arabic script for justification of running text in paragraphs, since I believe that in the end the requirements will be somewhat complex, and we are a long way from understanding them yet. I agree. > I guess we can say that inter-word turns off elongation, since that is the purpose of that value. I'm not sure we can say much yet, until we understand things better, of the relationship between distribute and elongation, since elongation seems to involve far too many, poorly understood, complications. Are you suggesting we should add “inter-word turns off elongation”, or it’s implied enough that the current text is fine? > The current text says "Justification adjusts spacing between each pair of adjacent visually-perceived characters", which may be sufficient to deal with elongations, since joined cursive letters don't have any spaces between them. I suppose it introduces a question about stretching of gaps inside a word where letters don't join to the left, though. This is about “distribute”, right? According to the spec, yes, non-joining letters will stretch gaps between characters. I’m not sure if it makes sense for cursive scripts, nor if there were any needs for “distribute” value in cursive scripts. Any suggestions here? /koji On May 28, 2014, at 0:56, Richard Ishida <ishida@w3.org> wrote: > I suspect we should say as little as possible for now in the spec about text elongation in Arabic script for justification of running text in paragraphs, since I believe that in the end the requirements will be somewhat complex, and we are a long way from understanding them yet. > > For example, it appears that some font types (eg. Ruqah style fonts) should not be used with elongation at all - all the justification is therefore done by manipulating space. Also, it's looking to me as if justification in, say, a naskh type font is likely to involve adjusting spacing a lot of the time, but may occasionally involve that plus some elongation when the spaces get too wide (eg. when a url is embedded in a narrow column). In other cases, however, a greater amount of elongation may be favoured in a naskh type font. Nastaliq font types, on the other hand, seem to rely more on elongation (and ligation and other shortening techniques) than naskh (since spaces are seen in a somewhat different light for nastaliq script). Also, btw, the rules for elongation are somewhat different in nastaliq than for naskh. > > Like detection of conjunct behaviour in indic scripts, the determination of the correct behaviour seems to be font dependent, and it's not clear to me how to handle that. Note that all this therefore also raises questions about how to do the right thing when a user changes the font by design or a font is substituted due to availability of fonts on a particular device). It suggests that maybe we need special controls for elongation that say something like: apply elongation if the font is X or Y only. > > (By the way, in passing, note that in Tasmeem it is possible to independently set parameters for the size of spaces inside a word (ie. where a letter doesn't join to the left) as well as outside a word.) > > I guess we can say that inter-word turns off elongation, since that is the purpose of that value. I'm not sure we can say much yet, until we understand things better, of the relationship between distribute and elongation, since elongation seems to involve far too many, poorly understood, complications. > > The current text says "Justification adjusts spacing between each pair of adjacent visually-perceived characters", which may be sufficient to deal with elongations, since joined cursive letters don't have any spaces between them. I suppose it introduces a question about stretching of gaps inside a word where letters don't join to the left, though. > > > > PS: For a discussion of nastaliq justification in Urdu see http://www.cle.org.pk/Publication/papers/2004/rule-based-expert-system.pdf > > >> Resent-Date: Fri, 09 May 2014 00:30:14 +0000 >> Resent-From: www-style@w3.org >> Date: Thu, 08 May 2014 17:29:44 -0700 >> From: fantasai <fantasai.lists@inkedblade.net> >> To: www-style@w3.org >> >> On 01/24/2014 10:30 AM, Phillips, Addison wrote: >>> State: >>> OPEN WG Comment >>> Product: >>> CSS3-text >>> Raised by: >>> Addison Phillips >>> Opened on: >>> 2014-01-23 >>> Description: >>> Section 7.3: Surprisingly there is no keyword specifically >>> associated >>> with kashida (other than "auto"). Shouldn't there be? >> >> There is no special keyword for it, just as there is no special keyword >> for standard Japanese justification. >> >>> Or is it a special case of "distribute"? >> >> No. It is most definitely not a special case of "distribute". >> >>> Some WG members [1] thought that "inter-word" implies turning >>> kashida off. >> >> This is true. It turns it off, because it only allows expansion >> at spaces. >> >> ~fantasai >> >> >> >> >

On 05/27/2014 08:56 AM, Richard Ishida wrote: > > I guess we can say that inter-word turns off elongation, since that is the purpose of that value. I'm not sure we can say > much yet, until we understand things better, of the relationship between distribute and elongation, since elongation seems to > involve far too many, poorly understood, complications. > > The current text says "Justification adjusts spacing between each pair of adjacent visually-perceived characters", which may > be sufficient to deal with elongations, since joined cursive letters don't have any spaces between them. I suppose it > introduces a question about stretching of gaps inside a word where letters don't join to the left, though. The current text also says: http://dev.w3.org/csswg/css-text/#justify-cursive which I think should cover the issue as well as we can at the moment. ~fantasai

On 04/23/2014 05:55 AM, James Clark wrote: > This is the most useful reference in English on Arabic text justification that I have come across: > > http://www.tug.org/TUGboat/tb27-2/tb87benatia.pdf Ran across this today, which is also somewhat interesting: http://www.cle.org.pk/Publication/Crulp_report/CR04_01E.pdf ~fantasai

On 24/01/2014 18:30, Phillips, Addison wrote: > State: > OPEN WG Comment > Product: > CSS3-text > Raised by: > Addison Phillips > Opened on: > 2014-01-23 > Description: > Section 7.3: Surprisingly there is no keyword specifically associated with kashida (other than "auto"). Shouldn't there be? Or is it a special case of "distribute"? Some WG members [1] thought that "inter-word" implies turning kashida off. > > [1] http://www.w3.org/2013/12/05-i18n-minutes.html The i18n WG discussed this at our last telecon, and decided that we are satisfied with the outcome. We will close this issue in our tracker. ri