[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: IE is dead
On Fri, 23 Jan 1998, Stanislav Malyshev a.k.a. Frodo wrote:
> On Fri, 23 Jan 1998, Peter L. Peres wrote:
>
> >The best non-english software has always originated in non-english
> >speaking countries ;) Do not expect too much from either competitor. They
> >are too busy fighting each other to pay attention to small details, and
> >small countries. Such as Israel for example. Non-english software gets
> Well, competition didn't prevent M$ from having some
> internationalization support in MSIE (far from perfect, but still at least
> something). Netscape, even in 4.0, has almost none, and which it has - is
> deadly broken. One that want Netscape to operate his native language (I do
> not speak for localized interface - it's too much to expect and I won't
> use it anyway, though many would) - has to apply various dirty hacks.
> Also, to take Russia - which is not so small country - Russian support in
> Netscape is not much better than Hebrew... It exists, unlike Hebrew
> support, but is both insufficient and broken. Do not know what about all
> others international ones that are inside Netscape - but seems they aren't
> better...
I don't know about Russia, but I know that they do not write from right to
left, and their characters fit into little boxes like Latin ones, so the
problems are with the integration and translation of data for the local
versions of the software. Incidentally, I'd like to know how Chinese and
Kanji are handled by our beloved L95 system. I suppose that MT can handle
it with Unicode but I kind of feel bad for the poor guys who use the other
kind of disaster...
The real problem is X11 which cannot 'switch versions' in 6 months like
the concurrence, and which is a system primarily designed for English, or
at least Latin alphabet languages. Porting X11 to Hebrew probably requires
an entire new release, starting from at least as low as the Athena widget
set, and all applications. Plus, it must remain compatible. This is a
titanic project but it can be done, the foundation is there.
> >really good when the author immigrates to the US and stops being paid with
> >little pieces of colored paper (many of which their issuers want back
> >badly, everywhere), and starts being paid *money* for his work.
> Example?
Linus Benedictus Torvalds
Dr. John Ousterhout (author of the tcl/tk system)
many others whose names slip my mind just right now.
>
> >Never. The poor guys probably had one hell of a time porting it from UNIX
> >to L95 in the first place. I believe that they have learned a few lessons
> >and won't make any more mistakes. No-one does development at a non-portable
> Do not think so. UNIX and Windows Communicators are quite
> different, IIRC, though similar on look. Also, some capabilities of
> Windows version are stripped from UNIX one (one example: UNIX versions has
> far less charsets in font selection box... there are many tiny examples).
This has to do with decisions made by the porting team. They are
political. Probably to do with what they think is the common denominator
of UNIX installations.
> You are thinking too good on them. Try reading Dilbert ;).
> >level anymore. There is a very high level description of the GUI and
> You bet almost everyone does. I may even name the One Platform they
> develop for. You may name it too, I suppose ;) ;((
I think that you are wrong about that. The level of abstraction required
for porting over so many platforms is so high, that it is probably an
in-house hacked version of a major high level GUI design package. It
almost certainly runs on UNIX primarily. If for no other reason, then
because of the history of the Netscape development. I suppose that if you
pick ANY box with the concurrence's OS and just put all the sources for
all the versions of Netscape Communicator on them, and run 2-3 compiler
sessions, it will not just crash, it will evaporate and leave no traces of
itself. It would even turn off the lights in the room before leaving for
the great void.
> >modules in relatively portable libraries. There are 3rd party tools that
> Been there, worked with that. Not bad, but still - you have to be strictly
> in the borders of that tool designer's mind - you cannot add some "native"
> tricks... And this is the temptation hard to avoid - and almost impossible
> competing with strictly-one-platform rival like Microsoft, which would
> certainly use every last bit of Windows API to pick you.
What you don't seem to see is, that no-one has the time to play games with
the APIs these days. The dirty trickery is, not releasing in-house details
that are crucial for the development of stable applications. That is where
gurus are employed to hack APIs to work *the way they should according to
their specs*. In top-down design, the API is used as per its
specification, not tinkered with 1 month before release (to induce bugs
and break everything).
> >assemble the contraption for various platforms. If any, the Linux port is
> >probably the hardest to maintain because there are no automatic commercial
> >tools for that under Linux. (no, I do not mean KDE or RCS, or makefiles).
> So, what exactly do you mean?
> Also, if they had such a scheme - why UNIX (and Linux, particullary)
> releases are so delayed? If they had true multi-platform development, at
> least betas release should be almost simultaneous... And certainly no
> delayed in months' terms.
There is no integrated translation tool from the GUI to the compiler level
for Linux as far as I know. This means, that they use the high
productivity tools for the other platforms, finish them and release them,
and then port for Linux by hand, probably from one of the cash cow
platforms (Is there a communicator for BSD ?). I strongly suppose that the
guys who port for Linux work on a voluntary basis for now, using in-house
tools and source, as the sales for Linux do not justify their pay. Not
yet.
Of course, this is my opinion from what I have read in the news and seen
done elsewhere. I may be very wrong. But I don't think so.
For a small enterprise it is possible to acquire a multi-platform tool-chain
(with a 5-seat license and no royalties) for about $35K. This is a
tool-chain that allows very high level GUI design, mock-up testing, and
generates code (C, C++) for several UNIXes, NT, L95, MAC. Then, on each
platform, an integration team connects the high level GUI generation
output code with the libraries that do the low level work (I/O, streams,
hardware control etc.). After this, there is alpha testing. This is the
way it works. And this is how the packages grow so large: no-one optimizes
the high level code with the low level code anymore. When there is serious
algorythmical work to be used, it is written independently by a separate
team and integrated with the GUI and low-level libraries.
> --
> frodo@sharat.co.il \/ There shall be counsels taken
> Stanislav Malyshev /\ Stronger than Morgul-spells
> phone +972-2-5369213 /\ JRRT LoTR.
> http://www.sharat.co.il/frodo/ whois:SM719-RIPE@whois.ripe.net
>
>
>
>
>
>
- References:
- Re: IE is dead
- From: "Stanislav Malyshev a.k.a. Frodo" <frodo@sharat.co.il>