Newbie FAQ #2: Where's the GUI?

Jonathan Gardner wrote:
> On Apr 1, 7:00 pm, vanekl <va...@acd.net> wrote: 
>
> I happen to believe that we write web apps today only to satisfy
> Windows users. See, Windows is so insecure you can't download and run
> software on your computer. There is no jail, no security measures in
> place to keep applications from messing with each other and with the
> OS.
>
> People feel comfortable running webapps because they seem to be more
> secure. But I know enough to say that they aren't secure, even with
> Firefox. They can't be made secure without the same measures it would
> take to make the OS secure.
>
> Firefox may be a better OS than Windows, but it is certainly lacking
> compared to what I know about Linux.
>
> Soon, it will no longer be the case that the majority of users are
> running Windows. When people get onto a decent platform with a decent
> security model, we'll all download and run apps without thinking about
> the security implications. It won't be possible for joe user to run
> any application that can destroy his machine or even hurt another app,
> no matter how hard he tries.
>
> (Using Windows, it drives me nuts that I have to get admin permission
> to install stupid programs. What in the world do they need permissions
> to mess with my OS for? I just want to put songs on to my iPod,
> dammit.)
>
> But more importantly, webapps are terribly restrictive. You can only
> do certain things, and you can't do those things well. Some of the
> ideas that webapps have forced onto us are good---throwbacks to the
> days of thin clients. But as an author of many webapps, I can't tell
> you how many times I have run into the limitations of the webapps--
> even with AJAX or what-have-you. They simply are not enough to get the
> job done. Some things, yes. But never everything, and I'm always
> sacrificing the user experience to fit the model of web app
> programming. This is not right. It's not the future of software.
> 
>
> Someday everyone will wake up and wonder why web browsers are running
> web apps when those same apps run on your desktop at a million times
> the speed with a million times more features and a million times the
> simplicity.
>
> One day, you'll have a document reader that reads any document
> anywhere on the internet with cross-referencing and you'll have a app
> launcher that can find and launch any application anywhere on the
> internet. The two will be different tools.
>
> We'll look back at the 1990's and 2000's, wondering why we wasted our
> time trying to do something useful on such a terribly restrictive
> platform.
> 
>
> HTML5 is simply (1+ HTML4). I don't like the direction of this,
> recursively speaking. Might as well write a linux emulator on Windows
> and have people download that instead of building it piecemeal.
>
> In short, I don't buy that we should all be writing web apps. Let's
> write real apps.

You're probably right, but this is what we got now. It's current
reality, it sucks, but it works, it's what people are actually willing
to bother using/"downloading" and we can build on it and hide some of
the nasty things .. (lisp is nice for this).

Maybe we can sneak in a proper Open Source VM plug-in eventually.
Something low-level, but with a proper common socket API. Maybe
JavaScript will get JIT support and caching of the resulting binaries.
Maybe the browser will get a real "drawing context" or "canvas"
eventually (because of the point above?). I don't know, but there seem
to be some activity indicating that we're heading in these directions.

--
Lars Rune Nøstdal
http://nostdal.org/

Jonathan Gardner wrote:
> But more importantly, webapps are terribly restrictive. You can only
> do certain things, and you can't do those things well. Some of the
> ideas that webapps have forced onto us are good---throwbacks to the
> days of thin clients. But as an author of many webapps, I can't tell
> you how many times I have run into the limitations of the webapps--
> even with AJAX or what-have-you. They simply are not enough to get the
> job done. Some things, yes. But never everything, and I'm always
> sacrificing the user experience to fit the model of web app
> programming. This is not right. It's not the future of software.
...
> In short, I don't buy that we should all be writing web apps. Let's
> write real apps.

Thanks for this great post.  I could not agree more.  Since Ken* joined
the web app band wagon, I have been afraid I'd be the only one who
believed in good old desktop apps.

Every time I log on to GMail (which a lot of people call fast) I get
annoyed by how slow it is -- compared to my Opera 9.50.  Finally we live
in an age where new software runs nice and fast on my four-year-old
laptop ... and then I should sacrifice all that for tcp/ip latencies?

I will go on working with cells-gtk3 and enjoy the seamless integration
of cairo (yep, finally it's done.)** and soon OpenGL.  What was the
equivalent of that in ajax land again?

HTH,
Peter


----
* Hell Ken, you wrote cello.  *All* the bells and whistles.  I still
can't believe you're into web apps now ...

** Available from me upon request.  I will publish it on my blog next
w, hopefully.

Jonathan Gardner wrote:
> wxCL certainly looks enticing. I'll give it a go when I give up on Qt.

Oh, btw, I remember you were looking for internal advocates.  So, I love
cells-gtk.  The port to cells3 runs fairly stable by now, I have cairo,
soon hopefully OpenGL -- and I like the looks of GTK.

IMNSHO, if you're looking for something , it's got to be celtk or
cells-gtk.

Peter

> [Snip]
> Thanks for this great post.  I could not agree more.  Since Ken* joined
> the web app band wagon, I have been afraid I'd be the only one who
> believed in good old desktop apps.
>
> Every time I log on to GMail (which a lot of people call fast) I get
> annoyed by how slow it is -- compared to my Opera 9.50.  Finally we live
> in an age where new software runs nice and fast on my four-year-old
> laptop ... and then I should sacrifice all that for tcp/ip latencies?

Hmm. I think webapps and desktop apps aim at different objectives.
Webapps shine when you have to provide a simple, consistent interface
across operating systems (and perhaps cell phones or other devices).
Desktop apps work best as specialized, powerful, monolithic (more or
less) systems. I don't use gmail for its speed, I use it because I can
access my mailbox from virtually anywhere. Viceversa, a web-based
Emacs (for example) wouldn't make much sense.
Surely lately there's an excessive hype on web applications, not only
in the Lisp world; still, people use desktop applications for almost
everything except some trivial tasks, or for things which are
inherently distributed like social networking, online games and the
like.

> I will go on working with cells-gtk3 and enjoy the seamless integration
> of cairo (yep, finally it's done.)** and soon OpenGL.  What was the
> equivalent of that in ajax land again?

I think the future will bring the convergence of web and desktop apps.
(in the meantime, kudos for cells-gtk3!). HTML and Javascript were
absolutely not thought as tools for building user interfaces. Java
applets could have been a step in the right direction, if it wasn't
for their limitations (too many security restrictions, too slow, ...).
There's a google SoC project on the integration of Lisp in the browser
(yes, I know about Kamen Lisp, but never got it to work). Maybe *that*
is the future?

Cheers,
Alessio

Peter Hildebrandt wrote:
> Jonathan Gardner wrote:
> 
>
>
> Oh, btw, I remember you were looking for internal advocates.

Oh, right. Well I would say something nice about AllegroCL but Edi won't
let me, and I would say something nice about Cells but /no one/ will let
me. Meanwhile my latest user (I am losing count! But then I only have
ten fingers...) says I am undercelling Cells, so I think I have found
the perfect sweet spot, no one is happy.

kenny

ps. I have been feeling bad about Cello having so few widgets but it
just occurred to me that this is Lisp, no one would use them anyway.
Hmmm.... sure eliminates a lot of documentation: you get the tcl/tk doc
for the event stream doc, opengl for the rendering doc, snack or openal
for the audio doc... hmmm... k

--
http://smuglispweeny.blogspot.com/
http://www.theoryyalgebra.com/

"In the morning, hear the Way;
in the evening, die content!"
-- Confucius

Alessio Stalla wrote: 
>
> Hmm. I think webapps and desktop apps aim at different objectives.
> Webapps shine when you have to provide a simple, consistent interface
> across operating systems (and perhaps cell phones or other devices).
> Desktop apps work best as specialized, powerful, monolithic (more or
> less) systems. I don't use gmail for its speed, I use it because I can
> access my mailbox from virtually anywhere. Viceversa, a web-based
> Emacs (for example) wouldn't make much sense.

Yep, I just wrote something similar elsewhere in this thread.  My
original post was just the (ridiculously exaggerated) answer to Ken's
plan of porting a Cello app to ajax.  I fully agree that there are
plenty of situations in which a web app makes perfect sense (eg.
administrative interfaces)

> Surely lately there's an excessive hype on web applications, not only
> in the Lisp world;

It is like that with everything, isn't it?  Once there is a shiny new
hammer, every problem becomes a nail again.

> I think the future will bring the convergence of web and desktop apps.
> (in the meantime, kudos for cells-gtk3!).

kudos go to Ken.  He did all the heavy lifting.  My share in the cells3
port is nothing but some random testing.

> HTML and Javascript were
> absolutely not thought as tools for building user interfaces.

Yep, that's my concern here, too.  If I wish to do everything with a web
app, than I'd like to have a platform designed for this use case
(starting with a browser which actually *is* an OS).

> Java
> applets could have been a step in the right direction, if it wasn't
> for their limitations (too many security restrictions, too slow, ...).

... and you forgot the main draw back:  Too much java!

> There's a google SoC project on the integration of Lisp in the browser
> (yes, I know about Kamen Lisp, but never got it to work). Maybe *that*
> is the future?

Goes in the right direction.  If someone somehow manages to make this
lisp somehow provide the same gui bindings as in -- say celtk ... wow --
you could simply run the same gui code in the browser or on the desktop.
Now I'd love that.

Cheers,
Peter


>
> Cheers,
> Alessio

Jonathan Gardner wrote:
snip
> I happen to believe that we write web apps today only to satisfy
> Windows users. See, Windows is so insecure you can't download and run
> software on your computer. There is no jail, no security measures in
> place to keep applications from messing with each other and with the
> OS.

Exaggerating much? No security? No permissions? If you're
suggesting that most of the windows security features are
bypassed by running as admin all the time, then I'd agree.

I believe the rise of the web app
has more to do with 1) piercing the corporate fire wall,
2) running software that doesn't require an install,
3) doesn't require administration, and 4) has a low
barrier to entry.

I realize that one data point does not an argument make,
but you might enjoy this story.

I work with a lot of women who are afraid of installing
software for some reason. I don't know why that is, but
it's just what I've noticed. They don't mind installing
shit^H^H^H^H a bunch of browser 3rd-party toolbars, but
they enter a catatonic state when they are forced to
install shrink-wrapped software. I can't explain it, but
it's just the way it is in the wild. A phenomenon that
may just be unique to my environs, but I doubt it.

Anyway, this one woman spends most of her time away from
the main office, and is such a klutz that she has been
given TWO computers, one located within the main office that
we install everything she needs to work with, and her laptop
that does nothing more than login to her office computer
remotely and use the apps installed on it. For some reason,
she actually prefers working this way, and the boss doesn't
care one way or another as long as she gets her work done.

What I'm trying to say is, there is a group of
people who have so little confidence in themselves to
install software on their computer, that they prefer
other people to set it up, even if that means the purchase
of twice the number of computers. I couldn't believe it
the first time I saw this, but it does go to show to
what extent some people will go to avoid having to
tinker with their computers.

Many people are afraid to install software on their
machines, and zero-install web apps solve that. Installing
software is scary. For the average user, security really
doesn't even factor into it much; users are treading in
territory that is foreign to them, and that can be
frightening.



> People feel comfortable running webapps because they seem to be more
> secure. But I know enough to say that they aren't secure, even with
> Firefox. They can't be made secure without the same measures it would
> take to make the OS secure.
>
> Firefox may be a better OS than Windows, but it is certainly lacking
> compared to what I know about Linux.

Firefox is a poor OS. It's strength is that it is a cross-platform,
standardized VM in which to run serialized apps. It's a layer on top
of the conventional OS layer, as I'm sure you're well aware, not a
substitute. Ever try to run more than one javascript script concurrently?
You can't, not within the same instance of FF. Again, it's a poor OS.

> Soon, it will no longer be the case that the majority of users are
> running Windows. When people get onto a decent platform with a decent
> security model, we'll all download and run apps without thinking about
> the security implications. It won't be possible for joe user to run
> any application that can destroy his machine or even hurt another app,
> no matter how hard he tries.

Sun tried this, and for a variety of reasons, it failed. The reason
why I hated it is because bandwidth was usually not sufficient to
achieve a satisfactory start-up time. I also hated Sun's UI choices,
but I could've gotten over that.


> (Using Windows, it drives me nuts that I have to get admin permission
> to install stupid programs. What in the world do they need permissions
> to mess with my OS for? I just want to put songs on to my iPod,
> dammit.)

That sounds to me like your music library directory permissions are
not set up right, but I'm just guessing. If you've got this set
up correctly, however, I agree with you, data files should not require
admin rights, in most circumstances.


> But more importantly, webapps are terribly restrictive. You can only
> do certain things, and you can't do those things well. Some of the
> ideas that webapps have forced onto us are good---throwbacks to the
> days of thin clients. But as an author of many webapps, I can't tell
> you how many times I have run into the limitations of the webapps--
> even with AJAX or what-have-you. They simply are not enough to get the
> job done. Some things, yes. But never everything, and I'm always
> sacrificing the user experience to fit the model of web app
> programming. This is not right. It's not the future of software.

Maybe a maturing of the web platform and a widening of the pipes
is all that is required?

For example, MIT's X has separated display from computation for decades.

To say that this is not the future of software is, to me, not
very convincing.

 
>
> Someday everyone will wake up and wonder why web browsers are running
> web apps when those same apps run on your desktop at a million times
> the speed with a million times more features and a million times the
> simplicity.

Response time isn't everything, nor is development simplicity, but they
certainly can be useful.

> One day, you'll have a document reader that reads any document
> anywhere on the internet with cross-referencing and you'll have a app
> launcher that can find and launch any application anywhere on the
> internet. The two will be different tools.
>
> We'll look back at the 1990's and 2000's, wondering why we wasted our
> time trying to do something useful on such a terribly restrictive
> platform.

I think the trend is to go where the users want to go, not where the
developers want to go.

Yes, things are in a primordial state now. Doesn't mean they will be
forever. To wit:
 
>
> HTML5 is simply (1+ HTML4).

More importantly, we are finally getting movement for a maturing of
this new platform.


> I don't like the direction of this,
> recursively speaking. Might as well write a linux emulator on Windows
> and have people download that instead of building it piecemeal.

If people want to run both Linux and Windows on the same machine, they
can do that now.
I'm currently running coLinux/Debian and Win XP on the same machine.
[This was the best way for me to get SBCL to run alongside Windows.]

> In short, I don't buy that we should all be writing web apps. Let's
> write real apps.

What are "real" apps? Didn't you use to partition your code
before the advent of the web so that UI wasn't mixed with core logic?

I long for the ol' days of programming, too, just because it
was easier to program. But I'm willing
to trade more difficult development and a slightly poorer user experience
for,

o  wider audience,
o  centrally-maintained apps with extremely fast update cycles,
o  better user monitoring/tracking/feedback, and
o  ability to tunnel under the corporate firewall.

We're just heading in opposite directions, and I think the trend
is for wider dispersement.

BTW, feel free to agree to disagree and we can leave it at that.
The dichotomy of apps we're discussing will co-exist for quite
some time, and I understand your current frustration.

Ken Tilton wrote:
>
>
> Peter Hildebrandt wrote:
> 
>
>
> Oh, right. Well I would say something nice about AllegroCL but Edi won't
> let me, and I would say something nice about Cells but /no one/ will let
> me. Meanwhile my latest user (I am losing count! But then I only have
> ten fingers...) says I am undercelling Cells, so I think I have found
> the perfect sweet spot, no one is happy.
>
> kenny
>
> ps. I have been feeling bad about Cello having so few widgets but it
> just occurred to me that this is Lisp, no one would use them anyway.
> Hmmm.... sure eliminates a lot of documentation: you get the tcl/tk doc
> for the event stream doc, opengl for the rendering doc, snack or openal
> for the audio doc... hmmm... k
>

I knew there was something. OK, I am re-assigning Jonathan to take over
the prime maintainer role on Cello.

hth, kenny

--
http://smuglispweeny.blogspot.com/
http://www.theoryyalgebra.com/

"In the morning, hear the Way;
in the evening, die content!"
-- Confucius

Alessio Stalla wrote: 
>
> Hmm. I think webapps and desktop apps aim at different objectives.
> Webapps shine when you have to provide a simple, consistent interface
> across operating systems (and perhaps cell phones or other devices).

There's nothing that requires the engine behind the webapps to actually
be on a distant machine. Everybody has a browser, and those browsers can
be remarkably consistent in what they display as the result of a bunch
of code. So sure it's lowest-common-denominator in some ways, but it
gets you away from a lot of OS and graphics-package-dependent issues
that could otherwise bog you down before you even got started.

paul

On Apr 3, 1:11 am, vanekl <va...@acd.net> wrote:
> OS.
>
> Exaggerating much? No security? No permissions? If you're
> suggesting that most of the windows security features are
> bypassed by running as admin all the time, then I'd agree.

I think that's a good summary.  My impression (based on reading at
some depth about NT, but not that much since) was that Windows has a
security model which puts all but quite recent Unices to shame, but
that people run with it effectively turned off almost all the time,
and most developers make that assumption)

It's amusing how hard it is to do things right.  On my mac, the user I
log in as is not an admin user (an admin user being, in fact, just one
who has permission to sudo root), and it's interesting how many things
break in niggling ways because of that.

>
> I believe the rise of the web app
> has more to do with 1) piercing the corporate fire wall,
> 2) running software that doesn't require an install,
> 3) doesn't require administration, and 4) has a low
> barrier to entry.

And likewise, I think that's a pretty good summary.  In particular I
think you have to understand how important the "zero admin" thing is.
There is pretty much nothing I would less rather do than spend time
updating and managing some application, and I'm a professional SA so I
know most of the tricks (of course, it might be *because* I'm an SA
that I'm so unwilling to waste yet more of my life on this).  And of
course, I also know what SAs cost, since I get to bill my time, and
it's pretty obvious that any organisation which is doing its sums
right will be busily trying to reduce their SA overhead, even if that
means slightly cruftier applications.