[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

useless, time-wasting reply. (Re: Debian Installer?)



From: Behdad Esfahbod <behdad at cs dot toronto dot edu> 

This is the last time I mention this: minibidi is just a hack.
If I see people keep sending minibidi patches to externel
projects, I'm afraid I have to get involved from the other side:
the external projects.


you shouldnt use a threat-like language, because all i can say
is "GO AHEAD", and i dare/defy you.

you were always against the idea of another implementation,
making the BiDi algorithm look like some sanctuary that only
the _MASTER_ has a key to. well guess what, its just an
algorithm that you were lucky to arrive at first. the fact that
you mention 6 implementations (a few more exist) and mine,
goes to show that its not that difficult or a big deal.

its just an algorithm that makes text _look_ nice.
i easily implemeted minibidi, and so can anyone else who gives it
a week of his time.

I tried to show that minibidi is not a real bidi implementation

looks real enough for PuTTY's and CUPS' developers.

so i guess you'll be freinds with them soon, eh ? "the other side"

by mentioning a few things it misses.  Just a short list out of
top of my head:

yeah right.

* Takes \Theta(N^2) time for an string of length N.

no it doesnt, its O(n)

* Does not implement rule L3 at all. 

i know about this, and as i said before PuTTY dropped the
composing chars, so there was need for it then. CUPS needs
it, (mainly because i implemented composing chars) so i'll look
into it.

* Is not portable: uses wchar_t. 

hehe, i can change wchar_t to anything in 30 seconds.
you're just pissed off and trying so hard.

* Just works for BMP (plane 0) (in fact dumps core for chars
  beyond BMP).


try to use common names here so the others can judge for themselves.
b means that minibidi only handles U+0000 to U+FFFF.
fribidi handles beyond FFFF probably because you had some free time or
you wanted to use it in dishing out other implementations.

* Has known bugs handling explicit marks.

so what ? i know, i'll get to it when i feel like it.

* Char type lookup is slow as hell.

do you really think i cant make it faster ? only behdad can ?!?!?!?!

* Is not tested AT ALL.

not true.

* Updating to later Unicode versions is not any easy.

its my job not yours. this is none of your business.

* Has quite a bad design: Table in header file.

see above.

* Does not expose all functionality of the bidi algorithm.

see above.

* Has no build system. Is not a library. 

declaring a new datatype and some extra crap is not mandatory.
rememeber its called *MINI*bidi.
also, see above.

* Does not follow the standard with respect to line-breaking,
  shaping, and bidi interaction.

dont see much difference in output or anyone complaining but you.

And note that this last one is really hard to implement, and is 

hard to implement !!!!!!!!
umm, let me guess, only fribidi implements it correctly.
the fact that you gave me a couple of hints when i was implementing
minibidi, doesnot make you better than me. you just saved me some
time.

why shaping is not in FriBidi yet... As I already have said lots

umm, is it too difficult for you ?? feel free to use minibidi's shaping code

of times: Bidi/Shaping is not done line by line...

i do it line by line and i dont see any complaints. except yours.

And I can express how important these points are, and how much
evil is it to duplicate code in different projects.  Right now we
have only 6 bidi implementations used widely (FriBidi, KDE,
Mozilla, OO.o, Java, ICU), and we've got enough headache with
them, now go on and send this ugly hack all over different
projects...  So, in short, is not conforming to standard in any
level.

headaches come from posts like yours. dont see any other sources.

your arguments are weak dude. and you tried to make them look
smart by adding some useless jargoon.

you calling minibidi an "ugly hack" also shows that arabs in general
are the _worst_ community (considering its size) that never unites,
and always destroyes itself from within. instead of reporting your
comments in shape of feedback you decided to do it this way, proving
that point.

minibidi might not be the BEST solution, but its a GOOD ENOUGH.
dont forget that its a 2 week project, while you've spent years on fribidi.
i guess i'd be bitter too if i spent years on minibidi and then someone
implemented a similar solution in 2 weeks.

also, dont forget that _YOU_ inspired minibidi. or should i refresh your
memory with some quotes from the archives ??

i made minibidi MIT because i cared about the arabic community. wether
its open source projects, or commercial projects, the _ARABS_ will benefit.
if you remember you didnt allow us to use Fribidi, and when i posted about my
implementation you came running and offering to change the licence and
granting a special licence to putty and all of that.

if you keep pushing me like this i'll have to leave CUPS and get back to
minibidi and make it better than Fribidi (very easy beleive me), and we
could have a nice little cold-war here, what do ya think ? and then we
would be typical arabs. fighting each other and leaving the main issues.
probably would take me 1-2 weeks, you on the other hand need more
time.

you see in this reply i didnt comment on fribidi, i went straight for the
author.

well, im out of flaming for now, but i bet i'll have some more when you
reply to this post.

- eagerly waiting.

ak.

_________________________________________________________________
MSN 8 helps eliminate e-mail viruses. Get 2 months FREE*. http://join.msn.com/?page=features/virus