[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Standard for plug-in modules for application software?
Eli Marmor wrote:
>
> 2. When you implement a BiDi support for a GUI toolkit, the effort
> required to implement the layout algorithm is about 1% (one
> percent) of the whole effort. If you spend so many discussions
> for this so little issue, I guess we are going to have years of
> discussions regarding to the whole effort.
I really don't think it's a 1% (much like 40% if u don't have any BiDi support
including Unicode) - but we won't argue about that.
> 3. The main motivation of this method (of having a "Standard for
> plug-in modules" for multi-lingual support), is to have a
> standard algorithm through differnt GUI toolkit and different
> platforms. So if we do it, it must be compatible with something
> valuable. At least Unicode (though I believe that in order to
> allow Windows users to migrate to Linux, it must be 100%
> compliant with MS BiDi algorithm). Otherwise, why wasting time
> for such a standard? Just to have a compatibility between Qt and
> GTK? I wish that even only one of them had a good Hebrew
> support...
Oh, they will. I Just spoke recently to some person in IBM in US who told me he
was in the group of writing Full Hebrew/Arabic support for Netscape (before they
were sold to AOL) and the finished to write and gave it to Netscape, so if
Netscape will release the code (as far as I know it's not on the CVS tree) - we
will have full BiDi support in GTK.
>
> 4. There is a direct correlation between the level of discussions/
> noise/blah-blah, and delays in implementation. Microsoft does not
> discuss standards etc., but implements them. On the other side,
> the UNIX camp was so bad in development and implementation, so
> the UNIX vendors preferred to spend so much time in defining more
> standards, and creating more groups, etc., and didn't have time
> for development. Believe me, I know the UNIX market for many
> years. I am tired of COSE / OSF / UNIX-International / ACE /
> X/Open / UNIX95 / POSIX / XPG / SVID / The-Open-Group / and about
> 40 (!) other groups and standards. These discussions killed UNIX.
> So PLEEEEEAASSE! don't repeat this fatal mistake of the UNIX,
> and don't waste so much time for gossiping and prattles.
> You can argue that these discussions are not at the expense of
> the development process; So I can witness, at least for myself,
> that the past discussions wasted so much time of me; If these
> discussions were avoided, I can promise you that I would already
> have a good Hebrew support for Linux.
Already?? How? share us..
I do agree with you about too much discussions, but if you'll look at a
"profile" of Linux-IL users, you may actully find that maybe 1-5% of them DO
write for Linux.
>
> 5. There is a direct correlation between standards and low quality.
> I can give you many examples, but instead of going far, just take
> PLS (mentioned above) and any toolkit which is based on it, and
> test its quality. For example, CTL crashes more frequently than
> anything else I know, when it doesn't crash - it smears/smashes
> characters on each other, and usually it just doesn't do
> anything. Not because PLS is bad, but because separating the
> process into two separate parts has a cost. For having a separate
> module, you pay by lower quality; You must decide if you are
> willing to pay this cost.
> The BiDi of MS, with all the jokes, is MUCH better than any other
> current BiDI. If you think that a BiDi which crashes so frequently
> is good for you, than a BiDi which doesn't crash but works well is
> excellent for you, so at least stop the jokes about Microsoft (I
> think we deserve a higher level of quality, so I can continue with
> jokes about Microsoft...).
Totally Agreed.
>
> 6. I know that Freshmeat is full with GTK and/or Qt based
> applications. But I don't see the current status, of an absolute
> domination of Motif, changes so much. Mostly from 90% to 80%.
> Don't expect any of the new ports (Oracle/Informix/DB2/CA-Ingres/
> Sybase/and dozens other commercial applications) to port from
> Motif to GTK or to Qt. Or even to Lesstif. I know them, they even
> can't migrate to Motif 2.0 or 2.1 (from 1.2.*). It is not easy,
> and trying to convince them to migrate to another toolkit may end
> up by them giving up the Linux port.
True.
> In any case, before running and developing Hebrew support for all
> the toolkits (GTK, QT, XForms, fltk, lesstif - to name only a
> few), it may be wiser to wait a little more and see who is the
> winner; I am quite sure, that at least in the front of GTK and Qt
> we are going to face some surprises in the near future (don't ask
> me who will win; I don't have any idea...).
Totally dis-agreed. In this unfortunate days, you find yourself in a situation
that every application is written at least twise (examples: kmp3 and gmp3, kicq
and gicq etc...).
Since the QT library REQUIRES to write only in C++, then writing hebrew support
for it is a must, since it's very easy to port it then to GTK (which u can write
for it with C, C++ etc...) - and these are the main libraries that are being
using today in Linux in the free applications (ofcourse - all other commercial
stuff is written in Motif - and the Motif Hebrew support - I really don't know
about it. If there was some sort of a way to replace a static Motif library with
a static Motif library with full Hebrew/BiDi support - that was cool, but I
don't know almost nothing about Motif).
>
> Summary: You should not want a standard plug-in module for layout. If
> you still want it, there is already one. And if you want to invent
> your own one, at least do it cleverly. In any case, please stop
> wasting time by endless gossiping. You just hurt the issue. If you
> want to help, just develop. Like Dov. Although I didn't test his work,
> it looks that he is doing very well. He doesn't make noise, but
> develops quietly.
And very cool stuff! (just fix the hebrew on the screenshot, it's embarrasing).
>
> If you still insist of wasting time, and are capable of speaking, you
> are welcome to call me (050-237338); It is accessible almost 24 hours
> at day. If you are not capable of speaking (deafs etc.), and only if
> you are not capable of speaking, you may write me, but remember: My
> e-mail address is not <linux-il@linux.org.il>, but
> <marmor@elmar.co.il> !!!
>
I applogize for putting this on the mailing list, but if could someone write a
full FAQ, I'm williing to put it on my server TONIGHT/ASAP and then if someone
would ask any hebrew related question - he would be turned to the URL..
Last thing: Although there were lots of talking about the Hebrew language issue
- something good came out of this - check it out:
* Erez Nir is writing support for the KDE/QT libraries (I installed it today,
although there are no real RTL features, it's cool to see the menu's in hebrew -
our secretary thought it was "weird Windows" :) )
* Dov is doing some GTK hebrew support (didn't try that one yet).
Keep up the good work fellas.
For any comment, please email me at: hetz@dream.co.il
Thanks
Hetz