Romanian layout with ș and ț in it is not among the options in the GUI

Confirming the bug. Should be an easy fix somewhere in the keyboard layouts settings files.

For a web page explaining why the cedilla is wrong, http://www.secarica.ro/html/s-uri_si_t-uri.html sure uses cedilla a lot in the text body. Talk about mixed messages. In fact, the only instances of the comma I could find in a quick browse-through were in the last paragraph about MacOS, which contains the text "Ș/ș şi Ț/ț". Note that "şi" is still written with the cedilla though. *shakes head*

It's true that the cedilla versions are used a lot. But that doesn't make it right.
As a fix, I switched my layout from "ro std" to "ro academic". The only difference is that the cedilla is replaced with the comma.
This is the correct layout (used by both Windows Vista and MacOS X) and it should be the default in Ubuntu.

The reason for using cedillas in the page explaining why the cedilla is wrong, is because the correct glyphs are missing from Windows systems prior to Vista version.

Unlike Firefox or Opera, Internet Explorer up to version 6 is unable to make substitution for missing glyphs, so in order for a Windows user to be able to read what I have written, cedillas were used instead.

The same apply for many other Romanian web pages as well and I am expecting this situation to remain unchanged for the next 1-2 years, maybe even more.

Cristi

Cristi,

from your comment it is not clear to me whether you think it's better for Ubuntu to swicth to commas by default or not.

Microsoft has released patches for some of it's fonts http://www.microsoft.com/downloads/details.aspx?FamilyID=0ec6f335-c3de-44c5-a13d-a1e7cea5ddea&DisplayLang=en
so XP should be covered for those with genuine windows boxes that can do WGA updates.
See this for comments on the pack: http://blogs.msdn.com/michkap/archive/2006/12/09/1245128.aspx

I've *heard* (I don't have the time to search for it, maybe someone else will clear it up later) that EU mandates that each country's character sets must be supported by the OS (I guess for OS' that they deploy in institutions).

And anyways, using the comma-based characters IS the proper way to solve the problem at hand. Every other major OS is doing the right thing.

so which would be the preferred romanian layout by default? comma or academic?

Academic.
It is the primary layout in the Romanian standard SR 13392:2004 and it matches the key labels on Romanian keyboards.

comma == academic
cedilla != academic

I'd say comma==academic is better. As I pointed out, Windows has updated it's fonts to support these characters. So displaying these characters on XP, Vista, OS X and Linux will not create problems.
More over, some of the Romanian translations in Launchpad contain cedilla based diacritics instead of the correct comma based ones. I think the number of such errors can be reduced if the default layout produces the correct characters. Many computer users that do use diacritics in writing do not even know about this issue (comma vs. cedilla) (I know many very "computer literate" (professors, university assistants, C.S. students, etc.) that had/still have no idea about it.)

Lucian:

comma != academic

I am talking about the layouts. Both of these provide the comma t and s instead of the cedilla ones.
I agree the correct characters need to be generated, but not sure which of these two keyboard layouts to make the default.

MihaI:

while it may match romanian keyboards, those are vastly outnumbered by us style keyboard in romania, so I am not sure it
would be practical to use academic vs comma.

What layouts do OS X and Vista set up by default for ro?

"ro academic" and "ro comma" are xkb layouts that should replace those currently used, "ro" and "ro std". See http://diacritice.sourceforge.net/tastaturi.html for more details.

right now there's std and winkeys
we definitely should add comma and academic to the list
should we remove any of the existing ones?

We should remove the layouts with cedillas from the graphical configuration utilities. Advanced users that want to use those layouts can set them using "setxkbmap". This is the situation now, only reversed, the comma layouts are hidden.
Regarding "academic" vs. "comma", I strongly believe we should follow the standards. SR 13392:2004 defines "academic" as primary standard. Moreover, the de facto standard (i.e. on Windows Vista) is also "academic". I understand that programmers might be better served by the "comma" layout, but "academic" is much better suited for the general public.

Maybe some further info might help:
- the Linguistic Institute of the Romanian Academy only defines a comma unde s and t, so this is the way to go for any new project; cedilla unde s and t simply does not exists, officially speaking
- the cedilla under s and t is there only beacuse back in 1987 the ISO 8859-2 (Latin 2) standard associated these two characters with the Romanian language
- one of the reason for the above was that no Unicode characters s and t with comma below existed at that time; these have been introduced later, with Unicode version 3.0
- for 8 bit representation, the only standard that suits the Romanian language is ISO 8859-16 (Latin 10), which only appeared in 2001
- the only character recommended by the Linguistic Institute of the Romanian Academy that is outside the ISO 8859-16 standard is a character for "dialog and white pause, of equal lenght, which is longer than the hyphen sign"; in the Romanian keyboard standard, this has been associated with the en-dash (U+2013)
- Microsoft has no support for correct Romanian language in 8 bit representation; their CP1250 codepage is based around ISO 8859-2; no CP based on ISO 8859-16 actually exists; however, Microsoft strongly discourages *any* non-Unicode approach, so most likely 8 bit support for correct Romanian language will never be implemented (which is not that bad, in my opinion)
- officially, the Romanian language is normally supported by ISO/IEC 10646-1:2003 (i.e. Unicode) standard and by ISO/IEC 8859-16:2001 (Latin 10) standard if 8 bit support is explicitly required
- the correct Romanian ortography is enforced by law; because of this, Microsoft was forced to add support for correct Romanian language; this is limited to Unicode implementations, because the 8 bit support is officially discontinued; to date, only limited support is provided for WinXP (four updated fonts, Arial, Times New Roman, Trebuchet and Verdana); on the other hand, the fonts traditionally used for UI are already correct since long time ago (Microsoft Sans Serif and Tahoma); Palatino Linotype is also always correct

Now: the ş and ţ with cedilla should be left there only for backward compatibility reasons with some older applications. New keyboard layouts containing cedillas are NOT recommended, but may be necessary in some circumstances.

I don't know the past situation on Linux systems. I will describe wat is today the situation on Windows Vista:
- the old Microsoft Windows Romanian keyboard layout, formerly called "Romanian", is still there for compatibility reasons; it has been renamed to "Romanian (Legacy)"; the layout uses ş and ţ with cedilla, as before, and is a QWERTZ layout, as before
- layout 1 from the Romanian keyboard standard has been implemented with the name "Romanian (Standard)"; this one is considered the main Romanian keyboard layout, because it provides access to the complete Romanian alphabet directly on first level; this layout was first introduced as a standard in 1998 and has been later revised in 2004; this layout uses *only* ș and ț with comma below
- layout 2 from the Romanian keyboard standard has been implemented with the name "Romanian (Programmers)"; this ons has bee...

After reading Mihai and Criti's comments I understant this better and I also think that the preferences UI should only show the comma variants instead of the cedilla ones.

So
 Romanian Standard (called academic in xkb)
 Romanian Programmers (called comma in xkb)
 Romanian Legacy(?) (Called winkeys in xkb)

current std and default from xkb, the ones with cedillas should not be shown in the GUI IMHO.

I still think the default should be Programmers because even if it does not provide the diacritics in the first level, it would cause less confusion, most keyboards here having US layout.

This is best solved by upstream, it's not an Ubuntu issue, so xkeyboard-config should be told after there's some consensus what the best
way to change the layout file (if at all - it may need to include keys for the diacritics used by minority languages in Ro) and then to change
the gnome keyboard preferences applet as well.

I took the liberty of subscribing Misu Moldovan since he was active in the past in getting the romanian layouts in xkb and probably has some insight as well.

https://bugs.freedesktop.org/show_bug.cgi?id=371

Jani, I'm glad you agree with us on the necessity of using commas. However, I disagree in regards to the default layout.
Romanian keyboards are indeed in minority right now, but I'm convinced the situation will change. After all, the standard was published in 2004 and Windows started using it only this year. The manufacturers are starting to adapt. I've seen Romanian keyboards delivered with Fujitsu-Siemens PCs in a lab at "Politehnica" University and I've read that all the Romanians working for the EU are given Romanian keyboards.
If someone choses a Romanian layout he expects it to be different from the US one. Regarding the majority of users, Windows is ubiquitous and it's default layout is the Standard layout. Thus, there will be no confusion for the users; in fact, the keyboards will function exactly as expected.
I don't see any reason why Ubuntu should diverge from SR 13392:2004 which, coincidentally, is also enforced by the Windows monopoly. I think Programmers is the default layout in xkb because the matter was decided by programmers but we should not forget that Ubuntu is an OS for everybody.

Mihai, I have no strong opinion on this.

But we shouldn't equate the choice between std and programmer layout with the choice of primarily pleasing programmers or 'regular human beings' ;) Most non-techies still have us keyboards.

I think the idea is to annoy as few users with each release.

On Mon, 06 Aug 2007 12:58:13 -0000, Jani Monoses wrote:

> Romanian Standard (called academic in xkb)
> Romanian Programmers (called comma in xkb)
> Romanian Legacy(?) (Called winkeys in xkb)

Please ask Mişu Moldovan for more info about the "academic" name. I
remember there was an issue with someone (Sorin Paliga) who claimed its
acceptance for the use of the name "academic" for keyboard layout.
Perhaps a simple request sent to Sorin Paliga may solve the dispute
(Sorin is a Mac user). His e-mail is
paliga AT bastral DOT ro
and/or
sorin_paliga AT mac DOT com
Maybe Mişu Moldovan can tell more on this.

> This is best solved by upstream, it's not an Ubuntu issue, so
> xkeyboard-config should be told after there's some consensus what the
> best way to change the layout file (if at all - it may need to
> include keys for the diacritics used by minority languages in Ro)

Please look at this documents:
http://www.secarica.ro/kbdro_help_romanian_standard_en.pdf
http://www.secarica.ro/kbdro_help_romanian_programmers_en.pdf
While the Romanian keyboard standard does not force which accented
characters should be generated by the use of the dead keys, the actual
implementation should take care of the official minorities. The above
documents claims to cover all those.

Another comment:
- decimal separator from layout 1 generates comma
- decimal separator from layout 2 generates dot
For each, AltGr+[decimal separator] generates the opposite.
Maybe this is a detail that counts in which default keyboard should be
offered at install time, for example when some initial IP are set up,
or something else. Or maybe not. But it is better to look at all
aspects ...

Cristi

--
Cristian Secară
http://www.secarica.ro/

I agree with Mihai that the default should be Romanian Standard. I don't see good reason to deviate from the established standard. On the contrary, that fact that there is a standard should make the decision even easier.
However, I do see good reasons for not making Romanian Programmers the default. As a user, if I'd want to change my keyboard layout to Romanian, I (and most likely your typical Romanian user) wouldn't expect the diacritics to be accessible on the 3rd level. My first reaction would be to think that something is broken if diacritics didn't start appearing upon hitting keys on the right-hand side of the keyboard. I would think that the Romanian Programmers layout would be a keyboard layout foreign to most Romanians.

Since I have never used the ro keyboard layout (except now in connection with this discussion) I'll leave it to others to decide.
But there is a difference between seasoned users who have set up and used a ro layout in the past on a US keyboard and those
who are new to computers and their keyboards (most that are sold now are still US) does not emit the codes that are on the keys.

I personally know nobody with a hw ro layout keyboard but maybe I have been living under a rock :)

The bug is not a bug, it's a feature, the Romanian layout works as intended. The behaviour in xkeyboard-config will change when the time will come, see below for details. Unfortunately there are two different problems discussed in this bug:

1) should the diacritics in the default layout be changed? (that means: scedilla ->scommabelow & tcedilla->tcommabelow)
2) should the default layout be changed from "Romanian Programmers" (secondary layout in the the SR 13392:2004 standard) to "Romanian" (primary layout in SR 13392:2004)?

1. I think the time has almost come for the first change. MacOS has had the "commabelow" diacritics in the RO layout for a long time, but the use of the Romanian layout from MacOS in practice is rather insignificant. MS has just introduced the correct diacritics in the primary layout of Romanian Vista, but Vista is still a long way from becoming the mainstream OS, most Romanian users that use diacritics are still using the default layout from XP, with the "cedilla" diacritics. Although there are fixes for XP users (the aforementioned patch for the output and the layouts from secarica.ro for the input), people are mostly not aware of them and get angry when they see broken diacritics from Vista or MacOS (or hypothetically Linux) users. I think it's important to wait for the number of Vista and MacOS users to reach a critical mass before making this change, in order to annoy as few users as possible. We are close. (The "we must be part of the change" argument doesn't really weight much, the Linux usage of the RO layout is really insignificant in the overall picture and I think MS should bear the pain and the blame for this transition).

2. I don't agree with the second change. There are historical reasons for the "Programmers" layout being the primary layout in X and xkeyboard-config, the main one being the lack of hardware Romanian keyboards. No, Jani, you weren't living under a rock, I safely assume that under 1% percent of the keyboards in use in Romania are localized (and what a shame that is, even for a country from the Balkans, Romania is the only one with this problem). Besides that, I happen to know a few users of the Romanian layout in X (OK, I've just counted them, they are eight, not including me). Except one, they all prefer the "Programmers" layout. The one that doesn't use the "Programmers" layout uses the "winkeys" layout (the QWERTZ one from XP) because he is stuck with a German keyboard. Granted, my ad-hoc poll is a very subjective one, but let's build a better statistics from asking the Romanian translators what layout they prefer and use. I bet 10:1 that it's the "Programmers" layout. So what would be the reason for changing the primary layout? Windows has it's default own layout (which in Vista is one from the standard), MacOS has it's own wacky layout, xkeyboard-config has it's own default layout (which is also included in the standard). What is the problem with that? People get angry when defaults are changed (as Manfred Pohler's change in the X layout has already proved).

I've just returned from a week-long trip, please excuse my late reply. Thank you for subscribing me, Jani.

1. When do you think the time will come for the change? How about when a law is passed requiring the use of the correct characters in public institutions? Well this law exists since May 2006 http://www.legi-internet.ro/index.php/LEGE_nr_183_din_16_mai_2006_p/242/0/
Is the ignorance of some Windows XP users more important than that?
2. Here a some reasons for changing the primary layout:
- obeying the law; see article 4 here http://www.legi-internet.ro/index.php/Ordin_MCTI_privind_utilizarea/257/0/
- following standards; primary layout in SR 13392:2004 pictured here http://diacritice.sourceforge.net/tastaturi.html
Microsoft changed the layout in order to comply with the law. You really don't see any problems with disregarding standards?

I think in real life de-facto standards are more important than laws or official standards but also think it would be better to have commas instead of cedilla sooner rather than later.

We need to know what the costs are at this moment. What does the 'ignorance of some of the XP users' actually cover?
Does a document or web page written with ș and ț show garbage instead of those charachers for most XP users? If so that is something to
consider.

I do not think we need to wait though until Vista is widespread and ș and ț are safe to use in Ubuntu. That is what we very much try to prevent with Ubuntu actually ;)

Thanks for your comments Mișu. Here are my thoughts with regards to the points you raised.

1. I personally have a hard time reconciling the idea that in order to get Linux fixed, we have to wait for Windows Vista to gain significant market share. I might find this easier to understand if there weren't any standard to follow, but there is. I don't use Windows because things such as diacritics are broken, now if Linux just replicates those issues, then I'm not sure why I should use Linux. Admittedly this isn't related to freedesktop.org, but Ubuntu's bug #1 has the goal of decreasing Windows market. Seems silly to me to have a goal of decreasing Windows market share whilst at the same time waiting for an increase in market share in order to fix other bugs.

2. See Mihai Captă's comments. I agree entirely.

Forgetting the issue of what should be the default layout, can we at least agree that we need to have the option available in Ubuntu UI that allows users to select the Romanian keyboard layouts that use the correct diacritics? Both Vista and Mac OS X allow me that luxury.

The existing xkeyboard-config layout follows the standard to the letter. As Cristi Secară said it already, the SR 13392:2004 doesn't force you to set one of the supported layouts as primary and the other one as secondary. There is nothing against the law either in the Romanian xkeyboard-config file, the cited law only forces some categories of public clerks to use the standard, which they could very well do using the existing xkeyboard-config layouts. Let's not burn in flames over these issues. If you really want correct diacritics and know the difference, you should issue "setxkbmap ro comma" or "setxkbmap ro academic" and get what you want. There is nothing to "fix", none of the included layouts are "broken"...

As I said already, I think the optimal time to switch will be reached after a critical mass of Vista users will start to use the new diacritics and people will start to adapt to it. Right now I mostly see the negative reactions from the existing XP user base to the early Vista switchers that use the new layouts with commabellow diacritics. From what I see around me, the people have display issues with Word documents, webpages, mails, instant messages that use the new diacritics because they haven't applied the patch for XP that fixes the fonts. Sometimes it's not their choice, they are users in a corporate environment that have centralized policies for applying patches and hotfixes. Some software packages know how to borrow the commabellow diacritics from fonts that have it (Mozilla based products do this) and the text is legible, although noticeably uglier. Some other software have no workarounds and simply display empty squares. That hurts.

I very much respect the fact that Ubuntu users are early switchers and always eager to try the bleeding edge. But please understand the deeper aspects of integrating technologies that are used in other operating systems too... I would have switched in my translation to commabellow diacritics years ago, but I know there are a lot of XP users that use GTK+ apps and they will have issues with displaying my "perfect" diacritics.

I agree that we are discussing two separate problems here. We should split the conversation in two separate bugs.

1. It's true that we don't know what the ignorance of some Windows XP users covers, but that doesn't change the fact that it's only ignorance. We can't make decisions based on that.
I did change my layout to academic and I even did it using GCconf so it plays nicely with gnome-keyboard-applet but that's not the point.
The bug reporter is correct: the default Romanian keyboard layout does have incorrect characters (and so do the other ones that can be accessed using GUIs). When a user chooses a Romanian layout he expects it to produce correct characters. It isn't stated anywhere that the layouts work otherwise.
Regarding the law, of course it doesn't force a private person or company to write correct Romanian, but shouldn't we use it as a base for our decisions? How are we furthering the advancement of Ubuntu if it's not even readily available to users working for the state?
I totally agree with Jani and ctomer regarding Vista's market share.
This whole cedilla vs. comma issue is much bigger than just correct display. It affect language aids, search, database and many more categories of programs and I'm sure you are aware of that. It has nothing to do with the bleeding edge. Besides, Ubuntu is build for regular users, not for early adopters.

2. You say that "the SR 13392:2004 doesn't force you to
set one of the supported layouts as primary and the other one as
secondary" and I agree. But aren't the layouts in the standard called exactly "primary" and "secondary"?
Also, the law I mentioned earlier specifically recommends the use of the primary layout. Is there any good reason to disregard that recommendation?

3. The translations are a different issue that I have already raised on the Ubuntu Romanian Translators mailing list at https://lists.ubuntu.com/mailman/listinfo/ubuntu-ro

Ok, then please close this bug, it is misleading, there are no "incorrect characters" in the Romanian keyboard layout. Better open one for changing the default layout in the Romanian xkeyboard-config file to the existing "academic" layout. But this should rather happen upstream, in the xkeyboard-config bugzilla. But I'm telling you beforehand, this will most likely happen 6 to 18 months from now.

However, Ubuntu (or Gnome?) may tweak their graphical tool to their liking, without any change to the existing xkeyboard-config file, which fully implements the standard.

I think you misread my last comment.
I think the bug is valid and the bug reporter is correct: the default Romanian keyboard layout has incorrect characters. When a user chooses a Romanian layout he expects it to produce correct characters. It isn't stated anywhere that the layout works otherwise.

I hope simple sentences will help a bit:

1) There are no incorrect characters in the RO layout in xkeyboard-config. If you do not care why most texts in Romanian are written with sceddilla and tcedilla, that's another issue. The RO implementation fully implements the existing standard and may be used to enforce the new legislation.

2) The bug reporter has a problem with the graphical tool used to changed layouts. The layout that he wants as default exists in the xkeyboard-config file but it is not exposed in the graphical tool used in Ubuntu. Again, this is not a problem in xkeyboad-config.

3) The default layout in the Romanian xkeyboard-config file will change when most people will be able to *read* the commabelow diacritics. I won't call them "clueless" or "ignorant" or whatever. ETA: 6 to 18 months.

4) But it will still be the secondary layout in the new standard until there are a significant number of Romanian hardware keyboards.

You already have everything you need to fix this "problem" in your graphical tool. You have all my support on the xkeyboard-config side, but I don't think I have to add anything to xkeyboard-config to make your graphical tool do what the bug reporter wants. I'm open to suggestions and discussions if things are relevant to xkeyboard-config.

Unless anyone has any major objections, I plan on changing the title and description to something that specifically relates to the keyboard options in the Ubuntu UI, or lack thereof. Either that, or close the bug and raise one that more accurately describes the problem?

This will allow separate bugs to be raised, if necessary, for other issues such as the default layout, whether there should be cedillas at all, translations or whatever.

reassign to GUI package, as the xkb rules files contains the necessary Romanina layouts (even if they could use better names)

Mișu, I know most texts in Romanian are written with cedillas but I fail to see how that makes the cedilla characters less incorrect. I don't understand what you are saying. Are the characters correct because they are widely used?

ctomer, I think you should add "default" to the title and change the "steps to reproduce" part of the description to mention "setxkbmap ro" instead of the Ubuntu GUI. That would make the bug perfectly adequate for xkeyboard-config.

On Wed, 08 Aug 2007 15:06:47 -0000, Mişu Moldovan wrote:

> 3) The default layout in the Romanian xkeyboard-config file will
> change when most people will be able to *read* the commabelow
> diacritics. I won't call them "clueless" or "ignorant" or whatever.
> ETA: 6 to 18 months.

How long it will take for a change "here" to appear "there" ?
I mean the propagation delay. When will be the end user affected by a
today's change [in Ubuntu, and/or else] ?

Cristi

--
Cristian Secară
http://www.secarica.ro/

Dear Mihai, in theory there is no difference between theory and practice. But, in practice, there is... Reductio ad absurdum: let's suppose there are some incorrect characters in the RO xkeyboard-config file, can you please take a look at it and show us how to correct them? I suppose not, we were mainly talking about changing the default layout, nobody found flaws in the current RO layouts and the characters described there. In theory, the cedilla diacritics are indeed incorrectly used for writing Romanian, in practice however some users are still unable to read or type the commabelow ones, so they still use the latin2 diacritics. Most of them cannot tell the difference and do not give a damn. Cedilla diacritics are still needed for compatibility reasons and they are correctly described in the RO xkeyboard-file.

Jani, the names have their history. "std" and "winkeys" are names that xkeyboard-config maintainers use to describe "standard" and "microsoft style" layouts in all their files. "comma" was a name picked by Marius Andreiana for the ASTIQ variant that implemented the "commabelow" diacritics for the first time and "academic" was a name coined by Sorin Paliga, a linguist and MacOS user that has also participated to some of the meetings for SR 13392:2004 and that has been suggested to me by Cristi Secărică, the spiritual father of the new standard. If someone thinks of better names for the "comma" and "academic" layout, I could change them. However, keep in mind that the "comma" layout will become "default" in the near future and "default" will probably become "cedilla".

Cristi, a change in xkeyboard-config takes a while to propagate. First there needs to be an official release from the xkeyboard-config maintainers and then the distributions will start to include the new layout included in the new official release. Bleeding edge distributions like Ubuntu or Fedora Core would take about six months, others would take even more: 12 or even 18 months (think Debian stable). I'd say that if I want to change something in the xkeyboard-config file I would have to think 6 months in advance.

Dear Mișu, we were only talking about changing the default layout.
In practice, some users will never be able to read the comma characters (Windows < XP).
In practice, some users will always need to write using the cedilla characters for reasons of compatibility with legacy software. That is the only reasons why those layouts exist.
Cristian Secară explains this very well on his site and gives a warning to those who want to install the cedilla layouts on Windows Vista
"Atenţie ! Nerecomandat ! Instalaţi doar dacă ştiţi că vă este absolut necesar !"
that means
"Attention ! Not recommended ! Install only if you know it is absolutely necessary !"
I'm not saying you should remove the cedilla layouts from xkeyboard-config like Vista did.
All I'm saying is that you can't have such a layout as default. You can't have people unknowingly write texts using non-existing characters. How are people supposed to know that using the default Romanian layout produces incorrect characters? setxkbmap doesn't even print a warning message.
You say that people "cannot tell the difference and do not give a damn". Well, we know the difference and give a damn.

Mișu, we all agree the xkeyboard-config files have all the ro layouts there and setxkbmap can be used to choose any of them. (they may not have all the diacritics for minorities as required by the law, I am not sure)

We need those exposed in the GUI as well so anyone can choose them, that's why it has been reassigned to the appropriate package, I think there's no disagreement there either.

The names are only descriptive for those who know what comma and academic stand for in this context. The ones mentioned by Cristi
being used by Vista sound semore nsible to me. So we could have Romanian (now academic) and Romanian Programmers (comma) and the same ones with cedilla added to the name as well as keeping winkeys (instead of Legacy).

Does the name imply which is the default? If _now_ we changed these names which make sense in the long term, can we still make the cedilla version the default?

As for changes being prepared in advance that's not necessary at all, we can easily add a patch to xkeyboard-config in Ubuntu anytime, as long as there is consensus about what needs changing. So the obstacle is only deciding whether a change is needed and if it is needed now
no requirement to wait for upstream release if we think something should be done now. We should be as close to upstream of course, but not block on it.

having just seen what ș and ț are rendered in XP I tend to agree it is not a 100% good idea to use those by default :(
still, if those correct characters are not getting used the situation will persist indefinitely.

Let's not forget there's a patch from Microsoft that adds support for ș and ț in Windows XP.

On Wed, 08 Aug 2007 22:59:01 -0000, Mihai Capotă wrote:

> Let's not forget there's a patch from Microsoft that adds support for
> ș and ț in Windows XP.

Yes, but only for displaying ș and ț, not for generating ș and ț from
keyboard (it is basically a font update, nothing more).
Although a good thing, the update is limited to the most used fonts:
Arial and Times New Roman (for printed documents), Verdana (for web)
and Trebuchet (they said this is for MS Office, but I know that this
also solves a titlebar issue when the user selected desktop appearance
is Windows XP instead of Windows Classic).

Another "problem" is that the font update download *) (not the actual
install) requires genuine validation check, which I doubt it will pass
on many home users. In my opinion this will affect the reading on
web-based forums, but less the e-mail, because most home users
appears to be Yahoo mail users **). The standard Yahoo mail
interface generates broken encoding declaration anyway (it is fixed to
ISO-8859-1) and will ignore the incoming mail encoding declaration
anyway (it treates all incoming mail as ISO-8859-1), so most Yahoo mail
users will see garbage, regardless the cedila or comma story ***).

*)
http://www.microsoft.com/downloads/details.aspx?FamilyID=0ec6f335-c3de-44c5-a13d-a1e7cea5ddea&DisplayLang=en
**) merely my own supposition; comments on this may be required
***) this may change if their new beta interface will become widely
used, which I doubt

Cristi

--
Cristian Secară
http://www.secarica.ro/

Cristi, Windows XP users can generate their own keyboard layouts or can download one of yours if they want to generate ș and ț.
Users that have the "problem" you talk about can easily go around it.
What does Yahoo mail have to do with this bug? As you said yourself, the encoding doesn't support cedilla characters either.

On Thu, 09 Aug 2007 07:17:43 -0000, Mihai Capotă wrote:

> What does Yahoo mail have to do with this bug? As you said
> yourself, the encoding doesn't support cedilla characters either.

I was only trying to identify (vaguely) the target users affected by
this discussion. Yahoo users (apparently a large Romanian community)
are not affected, I mean from this point of view it does not matter
which Romanian layout becomes the default one.

Cristi

--
Cristian Secară
http://www.secarica.ro/

So lets assume Academic is set to default in Ubuntu's next release. How many Windows XP user will be affected by this change? It would only be Romanian XP users who access content written by Romanian users of the very latest version of Ubuntu, right? That's not a significant number of users, is it?

The users affected may very well install the update from Microsoft even after they access text with the comma characters. They see the text is rendered incorrectly and they install the update. Problem solved. I've seen it happen numerous times amongst people I communicate with.

Mihai: XP users will probably not know where to get that patch, or they would have gotten it by now. I think they'll just get annoyed.
There's enough garbage fonts on various sites for them to assume something is broken at the sender/website.

Ctomer: depends on how many pepe will use the new Ubuntu, it may or not may be significant. You need to think it as a percentage of
Romanian users not of everyone. Otherwise we may claim this bug affects an even smaller fraction of users :)

Mișu (heh, a side effect of this thread is that I actually may start using diacritics :) : the bug is indeed in xkeyboard-config not the UI.
/etc/X11/xkb/rules/base.xml, part of xkb-data only lists the 3 romanian layouts, so both Debian installer (and Ubiquity) and the GNOME keyboard dialog will show those.
Do you agree this needs a change at xkeyboard-config level?

Mihai,

1) how do those users know they need an update to see the characters?
2) is this not avaliable only for those who have alegal copy of XP (a small percentage)

Well, I'm glad we finally agree that the whole issue is about the default layout in Ubuntu's graphical tool and that there are no "incorrect" characters in the xkeyboard-config file. Thank you for renaming and reassigning. I also think that the descriptive names in the graphical tool should be consistent with the ones in Vista and Secărică's layouts.

Mihai, you still have to realize that the "XP problem" is not something "easily" circumvented. Most XP users don't have either the know-how or the administrative rights to fix the display and input issues regarding commabelow diacritics. You need to be an administrator to install the MS hotfix and Secărică's layouts. I know plenty of knowledgeable people in corporate environments who don't have the necessary rights to do it. There is also the very common scenario in which the home user has the admin rights but has no idea how to fix the problem, which he/she thinks is not his/her problem. Cristi Secărică has also mentioned that you need to pass the MS genuine validation in order to install the hotfix, something that many Romanian XP installations will fail to do, given the very high piracy rate in our country. Also, in regards to ctomer's latest reply, "Romanian XP users who access content written by Romanian users of the very latest version of Ubuntu" may not be a "significant number of users" but they surely are a very significant *proportion* of the users that access that content.

Please consider the above when deciding what to set as default in Ubuntu. I know that the average clueless users are not nearly as vocal as the (not necessarily more knowledgeable) early adopters, but please take into consideration the technical facts which unfortunately translate into the sad reality of the currently very poor support for reading commabelow diacritics. I did my best to ease this for users of free software by extending the DejaVu font family of fonts with the necessary glyphs and by fully implementing the new keyboard standard in xkeyboard-config years ago, but I still think it's not the right time to change the defaults in the RO keyboard layout. Let's not share the pain of the early Romanian Vista adopters that are now seeking ways to write again with broken diacritics. I think Vista is doomed to replace XP as the dominant desktop/workstation OS in the not so distant future, so the whole issue will settle in time. It may well be that Ubuntu's no. 1 bug is taking over the world of computing but until then, please let's keep in touch with reality.

Jani, I don't have the /etc/X11/xkb/rules/base.xml file in Gentoo, it probably is part of the old X.Org xkbdata data (the Gentoo package of x11-misc/xkbdata conflicts with the installed x11-misc/xkeyboard-config package). The Debian installer and Ubiquity should use the new xkeyboard-config data.

On Thu, 09 Aug 2007 08:00:46 -0000, ctomer wrote:

> How many Windows XP user will be affected by this change? It would
> only be Romanian XP users who access content written by Romanian
> users of the very latest version of Ubuntu, right? That's not a
> significant number of users, is it?

Today, most computer users are Windows XP users :)

In my opinion the true compatibility problem comes merely from legacy
applictions, rather than from documents exchange between users.
What is the 8 bit versus Unicode situation on Linux ? How many
applications requires Latin 2 as mandatory ? What is the workaround in
such cases ?

Other than that, I tend to agree with Jani Monoses's comment "still, if
those correct characters are not getting used the situation will
persist indefinitely."

Cristi

--
Cristian Secară
http://www.secarica.ro/

Further investigation in the xkeyboard-layout package reveals the following in the rules/base.xml.in file:

    <layout>
      <configItem>
        <name>ro</name>
        <_shortDescription>Rou</_shortDescription>
        <_description>Romania</_description>
      </configItem>
      <variantList>
        <variant>
          <configItem>
            <name>std</name>
            <_description>Standard</_description>
          </configItem>
        </variant>
        <variant>
          <configItem>
            <name>winkeys</name>
            <_description>Winkeys</_description>
          </configItem>
        </variant>
      </variantList>
    </layout>
    <layout>

I'm not sure why this is, I'll probably fill a bug in the xkeyboard-config bugzilla. However, the problem is outside the RO xkeyboard-config file, so it will be up to the package maintainers to fix it.

Jani, I was talking about users that saw broken characters in emails or instant messages I sent them. They complained and I let them know there is a patch that fixes their problem.
I didn't know we were providing solutions for and adapting Ubuntu to users of unlicensed Windows XP, but if this is the case you should know Windows Genuine Advantage is easily broken and a quick googling will reveal how to avoid WGA checks.

Cristi, I totally agree. We must do something about this problem, not wait for Vista to take over the market.

right, so even if not in the romanian layout file, it is in a part of an xkeyboard-config supplied file that should be authored by romanians.
So that file needs the comma and academic layouts mentioned and they'll be picked up by any GUI that uses xkb.

Since there is going to be a change in xkb I think it's time to get change the names as well to have something sane in the long term.
Surprise for current users who are more knowledgable is more desirable than inflicting confusing legacy names on all users for years to come.

Mișu, I'm sorry to say this, but I don't agree. The issue is with xkeyboard-config's default layout as well.
We also don't seem to agree about the correctness of the characters: the cedilla ones are incorrect but we use them for a very clear reason. Please tell me where I'm wrong.

Mihai,

if I had started using diacritics and someone chatting with me had told me they're broken I would have not known there's a patch for XP.
Please stop ignoring the fact that most people have no idea what patches are, do not know about WGA etc. They just see the computer is doing yet another annoying thing.

I think we should move to standard layouts sooner rater than later, but by _being aware_ of the implications and risks and trying to mitigate them, not by waving them out of the way as if they didn't exist. Please stop saying XP users will know what to do. Mostly they won't. And additionally there's the admin rights thing Cristi bought up.

So we need to expose the comma layouts from xkb-data. That is necessary regardless of which will be the default.

Jani, I fully understand that people don't know much about computers and just want to use them.
This is precisely why I want the default layout to be correct, because people will never change it. People will be writing with incorrect characters until the default changes.

I don't think anybody is opposed to the exposing of the comma layouts.

Cristi, the situation regarding UTF-8 support is quite good in Linux. Most modern graphical apps have no problems. However, there are still console apps in use which do not support UTF-8 and maybe some ancient GTK+ 1.x packages. I'm not sure about the Linux console though.

Jani, I'm not really sure what that xml file is used for. I certainly haven't added the existing "std" and "winkeys" variants in it, they were probably filled in by one of the maintainers. There also seem to be plenty of variants in the symbols/* files that are not represented in that file. Regarding apps that graphically expose existing xkb variants to users, I know at least one case in which the GUI directly picks up the variants from the symbols/* files, KDE's control center does that. I will further investigate this issue and will probably report it in xkeyboard-config bugzilla.

Mihai, I'm not sure what points you disagree on now. It may well not be your fault, as this interface is not well suited for debates. However, I cannot see the end of our contradiction regarding the "correctness" of the diacritics in the RO xkeyboard-config file.

The general discussions over diacritics, layouts, UTF-8, xkeyboard-config etc. should probably take place somewhere else, we shouldn't pollute this bug report with endless debates. I would suggest the "Diacritice" mailing list, a place for people interested in Romanian localization of free software to discuss general localization issues: http://groups.google.ro/group/diacritice . That should also ease the communication as it is throughly suited for discussions.

Mișu, I agree. We should continue our debate on the "Diacritice" mailing list and return to this bug when we have reached a conclusion.

Hey, hey

I have a patch for the presentation issue. I used the names "Programmers", "Standard", "Programmers (cedilla)", "Standard (cedilla)" resulting in a dialog like the one in the attached image.

The brother Debian bug of this bug is 436993

http://bugs.debian.org/436993

The patch I made is available there.

the bug is not in gnome-control-center