Re: Borlands linux strategy ?

And if they deliver ASP.NET applications they aren't delivering solutions
to any browser? I don't care what server technology someone runs their
website
on when I go there. I just want it to work well.

Of course, but I apologize for not communicating my argument well. The
future sales model I'd like to see Borland help up move into involves
writing those servers you mention. With asp.net, we're still limited to
windows os servers.

I think Joel Spolsky said it quite elegantly about 5 1/2 years ago
http://www.joelonsoftware.com/printerFriendly/articles/fog0000000051.html:

So that's why FogBugz is delivered via a browser? That's the whole point of
SaaS, don't you see? In the future, even windows desktop development is no
longer the the most cost effective way to deliver many business apps,
because there's an even larger client-base, which is the browser. Joel's
argument isn't about Windows, but about the platform which provides the most
return for the development cost. What Joel's article does not take into
account is market maturity and competition levels.

The desktop os will lose relevance for the majority of business apps. So
will the server for that matter. If somebody wants me to deploy on his Linux
box (like the whole country of China for instance), then I should be able to
at least consider a way to make that happen with my favorite dev tools.
MSFT's whole future of software development efforts depend on the notion
that every developer includes Windows as part of his deployment requirement
in some form or fashion.

MSFT has to move a lot of product to keep the lights on, so if some huge
emergning market does go with non-windows or that Windows on the web thing,
then provided MSFT is still interested in doing developer tools, it will
have to consider doing something like making a mono that works, and then
platform-agnostic development will suddenly become the most reasonable move
ever, because MSFT promoted it.

See, this is what I don't get. Who cares if it helps or hurts MS?

MS does, and to some degree or another at our expense. You're arguing that
Delphi as a win tool provides the best solution because the universe of
solutions are delivered via windows and, for all intents, the universe is
windows. I would have agreed with this even a year ago, but now I honestly
think this mentality is holding us back from more opportunity and more
appropriate, maybe even more innovative solutions. My point is, we
exclusively favor MSFT at our expense.

Targetting a specific platform is about serving your customers in the best
way possible. IMO, I choose to target a market where the vast majority
run
windows; I like that market. Hence I choose to use Delphi. I believe
Delphi
is the best tool for the job at hand. If my application doesn't make
sense
in the windows market, I'll use a different tool.

If I'm an ISV and I want
to sell into Solaris shops, it makes sense to use Java or C++ or whatever.

Now are you really serious with that statement, or did you just make it
because you knew you'd probably never develop for solaris? Look, arguing
about a direction for Delphi I guess is really not that important, but it is
the dev environment that we love, and it's just a selfish interest that I
have for Delphi to be multi-target.

I suspect there's quite a bit beyond the pure OO stuff like containers,
etc.
that isn't platform specific. But things like the synchronization
primatives,
networking libraries, etc. are. Sure they could make them cross platform
as they did in CLX, but it's not trivial. It took Borland quite a while
to build support for Kylix. And that was still x86.

True dat, but a handful of extremely bright part-timers with no monetary
incentive seem to have done fairly well with the exercise.

If it's processor optimizations, then fine,
good point, but that is processors, not OS's, and the Fastcode model
could work well here.

How well is that model working for FreePascal? ;)

How would the level of difficulty be any higher than what FPC has done in
the first place?

I see that as being exactly opposite from this. DB Servers can benefit a
lot from offering multiple platform deployment; that was one of the
reasons
why Oracle was historically successful. SQL Server is the only mainstream
DB that only runs on one platform AFAIK.

Sorry, I didn't state the point very well. I agree with you on this. I meant
development of db servers of course has to take into account the os file
systems and thread/memory models the run on, so some good tweaking has to
take place for optimal platform performance.

Given. The rumor is though, $500 with unlimited deployment starting
with version 3.1, which is in beta right now.

IMO, that's still too expensive for these types of things. If it includes
a designer and things like that, they might get away with it, but not as
a corporate solution, not a mainstream one.

Seriously? If I had the backend squared away, and a universal frontend to go
with it, I'd pay a lot more than that for unlimited deployment.

It's also incredible that AFAICS Tibco's GI only supports IE (at this
time).
But IMO, that's ok...it's a targetted solution that addresses over 90% of
the market. ;)

You know they'll fix that.

But what other rich client toolkits are you looking at? Flash, for
example,
can make a client-only claim. OpenLazlo for example supports this kind of
development, they do indeed have a server piece but AFAICS, it's
optional...and
all their stuff is free.

Yep, flash, but I believe js has a better chance at working interfaces on
devices. Could be wrong here. OpenLaszlo is ok, but the standalone
deployment still requires a backend piece, like a standalone intraweb
server. Pure js still seems to be the cleanest solution.

Scriptaculous supports AJAX and has pretty nice effects; supports Firefox
as well. It's not necessarily like Tibco GI's UI, because IMO it's more
weblike not application like. OpenRico is another one built on top of the
prototype.js library. And I suspect other javascript libraries like Dojo
and Mochikit will move towards addressing visual widgets very soon. All
of these guys are free; all standard browser based stuff, all client-side
toolkits/frameworks.

I'll check them out, thanks.

Not to take anything away from the FPC guys, but how many real-world
applications
are built with FPC? How many with Delphi? How much total money has FPC
put in the pockets of developers using FPC? How much money from Delphi?

Why even pose these questions? The market is new and immature. How many
people drive hybrid cars? I'm ok with Delphi being a windows hosted app, but
eventually it's got to produce a hybrid car.

I would guess almost every professional Delphi developer makes money from
"limiting" themselves to the Windows platform. That's what Delphi is, a
Windows development tool. I'm not saying that Delphi will continue to
make
money for those developers going forward, I just think it has a better
chance
to do so if it stays focused on Windows.


There's definitely something wrong with that....have you looked at
BizTalk?
;)

Yeah, I think MSFT Biztalk is a pretty cool app, but more importantly MSFT
has a huge sales force and a captive market to push it and to accomplish
buyin. You sell windows apps, don't you? Same thing.

Still need a UI layer. UI layers tend to be platform specific (due to
user
acceptance) or Browser based (which one can argue is a platform itself).

Exactly! The universal ui platform running on the web os. Does the end user
ask what the backend is? Hell no, as long as it delivers the functionality.

No, actually I don't think that. I'm just not against dining at Bill
Gates'
table.

And you're paying for a seat.

I've seen many companies try to be everything to everyone and they
fail....miserably.
Remember when Borland tried to do office applications? ;) Or Corel try
to do Linux? IBM with OS/2?

"jack of all trades, master of none." That's not what I want Delphi's
moto
to be. I'd rather have it be a master of one.

Writing an application that consists of a backend that runs on more than one
os, and whose interface can be delivered via a browser is of course, not the
same thing.

James


.

Relevant Pages

  • Re: Remote info about computer
    ... nmap will try to guess the OS from the TCP/IP ... machine (I'm running nmap on the Linux server): ... Windows Me or Windows 2000 RC1 through ... Browser, Domain Master Browser ...
    (microsoft.public.security)
  • Re: IIS 5.0 NT Authentication
    ... How IIS Authenticates Browser Clients ... Windows NT Challenge/Response - The server requests the user to log on. ...
    (microsoft.public.inetserver.iis.security)
  • Re: Permissions Question
    ... What browser is the user using? ... If IE, ensure that windows integrated ... Microsoft MVP - Sharepoint Portal Server ...
    (microsoft.public.sharepoint.portalserver)
  • Re: The page cannot be displayed.....DNS Error
    ... > Operating System is Windows 2000 Professional ... > Browser is Internet Explorer: ... Cannot find server or DNS Error ...
    (microsoft.public.windows.inetexplorer.ie6.setup)
  • Re: Do You Need NetBEUI for XP Pro/Win98SE to Share?
    ... >Windows NT runs the browser, the client, and the server as services. ... so it uses anonymous access to any server it enumerates. ... >Computer or Network Server? ...
    (microsoft.public.windowsxp.network_web)