[Gllug] Application development across all platforms

[Darran D. Rimron-Molloy]

> It is currently written in VB with access holding the data.
>
> I want to web enable it.

Choice:
	a) Hack the VB to speak to the new database format and allow multiple
client types; or
	b) Write it in HTML, and everyone uses the HTML interface.

IMO) a is nice, native apps, et al but an arse to maintain when you want to
start adding features to the application.

> The data I plan to stor in a SQL database such as Postgres or Oracle.

I would use MySQL but I know others on this list who would argue
otherwise.... It's a personal taste issue for me.

> The difficulty I have it that the interface is relatively complex and I
> do not think would migrate to HTML easily.

I disagree, from the screen shots you present there, as long as you can
"force" some simple contraints on your user (must have HTML3.2+iFrames @
800x600 or similar) then you should be ok.

> I have seen the interface of Microsofts webmail which beautifully
> constructs the outlook interface in a browser. (It works on platforms
> other than Microsoft too ;-) Hence it is possible to do complex stuff.

And it works :o)

> The other alternative is do the interface in java.

Which has a large numbers of issues with it, JVM versions, Swing plugins,
etc. all a big mess, Java ain't what it used to be.

> I have also been looking at QT too as I read an article describing the
> coding of Hancom office in QT.  Allegedly it only took 30 minutes to
> alter the windows code to be able to recompile for Linux.

IMO, Bollocks. Sure, QT is supposed to be this WONDER-Widgit that anything
can be done in less time than it takes to crash a windows box, but I am a
cynic to that idea, always have been. LARGE portions of your code is going
to have to be ANSI C, with just the UI coded in some propriatory widgit set
if you want to get porting speeds even CLOSE to that.... VB is _not_ the
candidate for trying to prove this theory on.

> What we could
> then do is release multiple interfaces, one for Mac, one for Linux and
> one for Windows.  It is not really web enabled though!

Argeed, and can cause a whole host of other compatability/code maintainance
problems.

> As I shall not be programming this but collating alternatives and then
> proposing a solution I really would like input from the gathered mass
> :-)

Personally, I'd suggest a PHP/ePerl/[My|Postgres]SQL solution, redesign some
of the "More complex" portions of the interface to make it more accessable
from a web front-end -- sure it's nice to "spoof" the original, but at what
cost in development and userbility - I personally think that Outlook for the
Web would have been better if MS had tried less to clone the Win32
counterpart.

The other, slightly messy, complex and down-right hackery option is to write
it in an application server, port the whole darn show over to something
small and fast and native (C, obviously, springs to mind) either [x]inetd
spawned or forking using sopmething like libvnc (or alternative remote frame
buffer library) for clients - this is now entering the realm of real ASP
type stuff.

The overhead is, of course, is bandwidth, but with some clever coding, I
don't think that should present any sort of problem....

The application then has the EXACT same interface, is completely
net-enabled, can emulate the VB code, to the pixel, should you desire, and
can run on almost anything with a TCP connection, DOS, Windows, MacOS, *nix
[X11, Framebuffer and vanilla raw hardware], Amiga, WinCE, PalmOS,
blahblahblah, the list is HUGE - and, of course, because the VNC protocol is
open source, easy and designed for this sort of thing you can port it to
almost anything (completely ANSI C compliant code, then your application
server will run on almost anything too....) and can use ANY native API
available upon the host server, scratch disks, SMH, IPC, blahblahblah - alot
of possabilities here....

Prrof of concept can be produced, if you REALLY want, I've got some (closed
source) demos of stuff I put together for a customer about 18months ago, and
never really had the time to push any further....

	-Darran



-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug