qooxdoo trouble

[Kenneth Tilton]

Weird radio group behavior when one used a radio item that was a complex
widget itself. Dig dig dig... aha! I had noticed they had cocked things
up by not letting the complex item have a "model" property, and guess
where the problem originated?

The neat thing is I could then go here:

http://demo.qooxdoo.org/current/playground/#

Recreate the problem, and then post a link to it as an SPR!:

http://tinyurl.com/296hdjg

Click away from the complex item and then try to reselect it. No dice.

I then added the model property in my code base and got on with pretty
much finishing (in about six weeks full-time) the port of a desktop
application to the web:

http://teamalgebra.com/

Those six weeks included a refresher on qooxdoo as well as some new bits
and even worse, making friends with jsMath. Not that jsMath is not a
great, easy-to-use library, but I am using it more dynamically than it
is intended to be so ... actually, some work remains as you'll see from
occasional misplaced math or mis-sized containers.

Anyway, should be a much nicer year in Algebra for kids everywhere.

Thx again to the folks in comp.lang.lisp who turned me on to qooxdoo and
jsMath, and the folks in comp.lang.javascript who did absolutely nothing
to help except entertain me during the rough bits.

kt

 

[Kenneth Tilton]

re the "blank" keys issue Mr. Streater reported, I have had another such
report with screenshot and now notice a telltale pixel at the very top
of the buttons. This is my unresolved issue with getting positioning
info from jsMath so I can accurately position those things within
qooxdoo widgets: the character is there, it's just positioned outside
the widget.

Hmmm. Maybe this is a qooxdoo issue. They have an Html widget for raw
HTML. Seems to me it should accept responsibility for displaying all the
HTML absent any misguided directives from moi-self.

dig dig dig....

kt

 

[David Mark]

Weird radio group behavior when one used a radio item that was a complex
widget itself.

I could guess what that means, but I don't have time to speculate
about qooxdoo.

Dig dig dig... aha! I had noticed they had cocked things
up by not letting the complex item have a "model" property, and guess
where the problem originated?

See above.

The neat thing is I could then go here:

   http://demo.qooxdoo.org/current/playground/#

I will guess that - whatever that is - I wouldn't find it "neat".

Recreate the problem, and then post a link to it as an SPR!:

   http://tinyurl.com/296hdjg
Great.


Click away from the complex item and then try to reselect it. No dice.

You crapped out. Hate to say I told you so...

I then added the model property in my code base and got on with pretty
much finishing (in about six weeks full-time) the port of a desktop
application to the web:

Another patch? Upgrading that behemoth is sure to be that much more
fun.

   http://teamalgebra.com/

I've seen enough of that, thank you.

Those six weeks included a refresher on qooxdoo as well as some new bits
and even worse, making friends with jsMath.

With friends like that...

Not that jsMath is not a
great, easy-to-use library,

I established that it is not.

but I am using it more dynamically than it
is intended to be so ... actually, some work remains as you'll see from
occasional misplaced math or mis-sized containers.

No I won't.

Anyway, should be a much nicer year in Algebra for kids everywhere.

I doubt it.

Thx again to the folks in comp.lang.lisp who turned me on to qooxdoo and
jsMath,

LOL. Why would you ask LISP-ers to recommend browser scripts? You
got what you deserved.

and the folks in comp.lang.javascript who did absolutely nothing
to help except entertain me during the rough bits.

Clearly you'll never learn.

 

[David Mark]

re the "blank" keys issue Mr. Streater reported, I have had another such
report with screenshot and now notice a telltale pixel at the very top
of the buttons.

Oops, trouble in "paradise".

This is my unresolved issue with getting positioning
info from jsMath so I can accurately position those things within
qooxdoo widgets: the character is there, it's just positioned outside
the widget.

Ah, that's a shame. If only you'd use real buttons.

Hmmm. Maybe this is a qooxdoo issue.

Very likely suspect.

They have an Html widget for raw
HTML.

Again with that. Why would you use an "HTML widget" to write HTML.
The very idea is mind-boggling. Do they also have a "CSS widget" and
a "JS widget"?

Seems to me it should accept responsibility for displaying all the
HTML absent any misguided directives from moi-self.

Seems to me that you have abdicated responsibility for tasks that you
don't know how to do to others who are also without a clue. That's
certainly misguided (but a popular approach in Web development).

dig dig dig....

You'll be in China before this is over.

And enough with the qooxdoo diary entries. They are obviously of no
interest to this group.

 

[Kenneth Tilton]

Oops, trouble in "paradise".

Yes, I am ruined. All is lost*. I thought this would make you happy,
which is why I shared it.

* Not.

Ah, that's a shame. If only you'd use real buttons.

Actually, I think this is exactly /why/ I am not writing my own html.
But to use jsMath I will have to make it work within the qooxdoo
framework. Pray for me.

Very likely suspect.

I'll be sure to report back (as if you had any doubt).

Again with that. Why would you use an "HTML widget" to write HTML.
The very idea is mind-boggling. Do they also have a "CSS widget" and
a "JS widget"?

What is mind-boggling is your density. Move over, osmium! Aside from
qooxdoo's speed and quality, its big win is that one need not do
html/css. Unless of course one wants to, in which case one trots out the
qx.ui.embed.Html widget.

My guess (based on inspection of jsMath-generated html is that, for
things with interesting vertical positioning issues such as fractions,
jsMath is positioning stuff at a negative vertical offset (if up is
negative in html-land) and the qooxdoo widget is not allowing that. But
I /am/ guessing.

Seems to me that you have abdicated responsibility for tasks that you
don't know how to do to others who are also without a clue. That's
certainly misguided (but a popular approach in Web development).

You say "abdicate", I say "standing on the shoulders". As for "without a
clue", they have done a lot more with JS than you have based on your two
sites. You might want to take a break from Usenet and write some more
code, maybe you could have me as a user some day.

You'll be in China before this is over.

And enough with the qooxdoo diary entries. They are obviously of no
interest to this group.

Sorry, I did not realize this was your newsgroup... hang on...

kt

 

[Kenneth Tilton]

I could guess what that means, but I don't have time to speculate
about qooxdoo.


See above.


I will guess that - whatever that is - I wouldn't find it "neat".


You crapped out. Hate to say I told you so...

I was too subtle. There was no trouble, I said that just to get your
hopes up. There was a bug in qooxdoo, I reported it and they fixed it a
few minutes later since the fix was a no-brainer.

It's called programming. You should try it some time.

Another patch? Upgrading that behemoth is sure to be that much more
fun.


I've seen enough of that, thank you.

Look again, there is more every week. Now you can level up (ask a
teenager what that means) through Algebra. Lots of other improvements,
too, though lots more to come. Not alpha yet, but approaching it.

With friends like that...


I established that it is not.

Aw, come on, he's brought TeX to the web, give it up for the guy. Maybe
you'll like the successor: http://www.mathjax.org/

No I won't.


I doubt it.

Doubt these: http://www.stuckonalgebra.com/fan_mail.html

Two letters from teachers seeking out my software more than ten years
after their last use because (a) they missed it and (b) they could find
nothing like it. Excerpt to die for:

"Of all of the software I have used in the evolution of school math and
computers, your Algebra I Homework Tutor was the most elegant and
practical."

Eat your heart out! btw, if you surf the qooxdoo site each page refresh
brings a new, similar encomium from their happy users: http://qooxdoo.org/

Got anything like that for your library?

LOL. Why would you ask LISP-ers to recommend browser scripts?

Lispers know everything; I would not buy a lawn mower without checking
with them.

kt

 

[David Mark]

Yes, I am ruined. All is lost*. I thought this would make you happy,
which is why I shared it.

Yes, well that's a shame. But if you think my smileys mean that I'm
happy about something, you are mistaken.

JFTR, my smiley's, frownies, winkies, etc. are usually meaningless
mockery. Filter them out and the full meaning of the prose should
still be clear to all but the uninitiated. However the lack of a
smiley after a joke or ironic comment is one of my favored devices.

I can't understand why the typical Usenet denizen is confused by my
stuff. I often get the "are you serious?!" reaction, when it should
be clear that I have serious contempt for anyone who takes Usenet
seriously.

But I digress.

* Not.

In all seriousness, I'd lose that device.

Actually, I think this is exactly /why/ I am not writing my own html.

So, because qooxdoo's faux buttons exhibit problems that standard HTML
buttons do not share, you decided to eschew HTML for qooxdoo. Odd
choice.

But to use jsMath I will have to make it work within the qooxdoo
framework.

Your chosen math library requires qooxdoo?

Pray for me.

Sure. Send me $100(US) per month and I'll put in a good word for you
with the volcano God.

I'll be sure to report back (as if you had any doubt).

I don't doubt it, but I wish you wouldn't. You are simply wasting
time and space.

What is mind-boggling is your density.
Destiny?

Move over, osmium!
Who?

Aside from
qooxdoo's speed and quality, its big win is that one need not do
html/css.

Right, well we've been over that. The losses just keep piling up, but
you are too deluded to see it.

Unless of course one wants to, in which case one trots out the
qx.ui.embed.Html widget.

In which case their app will break down and have to be destroyed.

My guess (based on inspection of jsMath-generated html is that, for
things with interesting vertical positioning issues such as fractions,
jsMath is positioning stuff at a negative vertical offset (if up is
negative in html-land) and the qooxdoo widget is not allowing that.

Sounds like a match made in heaven.

But I /am/ guessing.

The issue was never in doubt.

You say "abdicate", I say "standing on the shoulders".

....of midgets.

As for "without a
clue", they have done a lot more with JS than you have based on your two
sites.

You don't build sites with JS, Kenny. Oh wait...

You might want to take a break from Usenet and write some more
code, maybe you could have me as a user some day.

You really are out of it, aren't you?

Sorry, I did not realize this was your newsgroup... hang on...

There have been a lot of complaints from the tenants.

 

[John G Harris]

http://teamalgebra.com/

<snip>

Still very slow to load.

Training center, Numeric Fractions, Adding and Subtracting :
The numbers in the boxes are only partly visible - just the bottom
quarter of the bottom number.

Is this the library problem you were talking about ?

John

 

[Nisse Engström]

http://teamalgebra.com/

Apparently, I have a flaky keyboard...

When I type "g=mc2==" on the [unbookmarkable tab], I get
"G0MC200" (and the "2" also triggers a browser shortcut).
And there are heaps of other peculiarities on that site.


/Nisse

 

[Kenneth Tilton]

Apparently, I have a flaky keyboard...

Or unsupported browser or both, but...

When I type "g=mc2==" on the [unbookmarkable tab],

Yeah, I hate that. I suppose I could find a workaround but not sure it
matters to my app.

I get
"G0MC200" (and the "2" also triggers a browser shortcut).

Re the latter, yeah, I need to get that preventDefault back in there in
re the shortcut.

Anyway, I am making headway on the jsMath and can confirm the "mssing
keys" phenomenon. My test of "without installed TeX fonts" was flawed,
but that is sorted out now. I am actually developing now without TeX
fonts installed. Next release later today will work /unless/ TeX fonts
are installed (but I think I can make it work both ways with a quick
hack, deeper fix in a day or seven).

kt

 

[Kenneth Tilton]

<snip>

Still very slow to load.

I need numbers!! Recently frickin WordPad on MS took ten seconds to
bring up the "Save As..." dialog. In user mode I went <sigh>, in
software critique mode I would scream "Death to Microsoft!". Louder than
usual, I mean.

Also browsers and geographical locations and... to be honest, I
sometimes see slow loads, then I reset my browser and all is well. My
Lisp server might be being stupid.

Training center, Numeric Fractions, Adding and Subtracting :
The numbers in the boxes are only partly visible - just the bottom
quarter of the bottom number.

Is this the library problem you were talking about ?

Yep. Not sure if it is qooxdoo, jsMath, or me.

jsMath expects to be generating Html embedded in a pile of raw HTML
written by David Mark, not a qooxdoo qx.ui.embed.Html widget programmed
by me. Somewhere from to three of us (me, jsMath, and qooxdoo) is
fucking up.

I blame Mark.

kt

 

[David Mark]

Apparently, I have a flaky keyboard...

That's as maybe, but the culprit here is likely Kenny's flaky app.

When I type "g=mc2==" on the [unbookmarkable tab],

That I don't mind as I don't think tabbed interfaces should mimic
navigation. They should persist their state though (e.g. with
cookies, local storage, etc.)

I get
"G0MC200" (and the "2" also triggers a browser shortcut).
And there are heaps of other peculiarities on that site.

Not unsurprising and it will require debugging a meg of dubious JS to
track them down.

 

[David Mark]

Or unsupported browser or both, but...

How do you quantify what you "support" with a browser sniffing
framework like qooxdoo? Do you just take their word for it?

And I believe Nisse is using the latest Opera.

When I type "g=mc2==" on the [unbookmarkable tab],

Yeah, I hate that. I suppose I could find a workaround but not sure it
matters to my app.

I get
"G0MC200" (and the "2" also triggers a browser shortcut).

Re the latter, yeah, I need to get that preventDefault back in there in
re the shortcut.

Which may or may not be effective.

Anyway, I am making headway on the jsMath and can confirm the "mssing
keys" phenomenon. My test of "without installed TeX fonts" was flawed,
but that is sorted out now. I am actually developing now without TeX
fonts installed. Next release later today will work /unless/ TeX fonts
are installed (but I think I can make it work both ways with a quick
hack, deeper fix in a day or seven).

Or seventy, by which time qooxdoo will have released a new version to
"support" the latest browsers and all of your patches will have to be
redone.

You are on a treadmill, Kenny. You really should learn HTML, CSS and
JS before trying to write a Web application. Basically, you are
trying to do algebra before you've learned basic arithmetic.

 

[Kenneth Tilton]

Apparently, I have a flaky keyboard...

That's as maybe, but the culprit here is likely Kenny's flaky app.

When I type "g=mc2==" on the [unbookmarkable tab],

That I don't mind as I don't think tabbed interfaces should mimic
navigation. They should persist their state though (e.g. with
cookies, local storage, etc.)

I get
"G0MC200" (and the "2" also triggers a browser shortcut).
And there are heaps of other peculiarities on that site.

Not unsurprising and it will require debugging a meg of dubious JS to
track them down.

Let's see what the browser/engine is first. Team qooxdoo seems to have
run up the white flag on Opera key events. My investors (me) are
prepared to lose that market. Of course if Mr. Mark can provide a
/universal/ HTML API for key events I'll just load his code and embed my
client code that uses it in a qx.ui.embed.Html widget and be singing his
praises too.


kt

I waded thru more than a little qooxdoo code to determine they had left
the model property off the RadioGroupBox widget. Doing so was not the
end of the world. Sorry. k

 

[Kenneth Tilton]

//teamalgebra.com/[/url]
Apparently, I have a flaky keyboard...

Or unsupported browser or both, but...

How do you quantify what you "support" with a browser sniffing
framework like qooxdoo? Do you just take their word for it?

You lost me there. Are you saying their browser-sniffing can guess
wrong? On what percent of users? Hint.

And I believe Nisse is using the latest Opera.

Opera? A tragedy, yes? From qooxdoo:

"Opera issues
Opera is the most broken browser concerning keyboard events. It supports
enough to use the keyboard for navigation but complex key commands
cannot be detected.

All key combinations containing the Alt key cannot be detected.
The modifier flags are only set in keypress events. This makes
qx.client.Command in Opera unable to catch most interesting keyboard
shortcuts.
Opera sets only the keyCode property of the DOM event. Because of that,
it is not possible to decide if the code in keypress events is a key
code or a char code. Since both codes may overlap, collisions may occur.
In need for a consistent solution, the current key handler
implementation always prefers the special keys over the normal keys. For
example, “s” and “F5” both have a code of 115 in Opera. In this case
qooxdoo would prefer “F5” over the “s” key."

Which may or may not be effective.

Well, it might not be live yet, but I have code in the dev version now
that refuses to handle anything but webkit, gecko, and ie.

I'll keep plugging on this issue, but what happened was I tossed the
prevent handling at a low level then later tossed all handling of what
qooxdoo calls keypress. The former was OK, the latter should have been
to still do the preventdefault but no more (ie. no round-trip to server
which was ignoring those events anyway).

Or seventy, by which time qooxdoo will have released a new version to
"support" the latest browsers and all of your patches will have to be
redone.

Are you hoping it is seventy, or hoping I get it sorted out tonight?

You are on a treadmill, Kenny. You really should learn HTML, CSS and
JS before trying to write a Web application. Basically, you are
trying to do algebra before you've learned basic arithmetic.

One thing you do not seem to be coming up to speed on as quickly as I
thought you might is that I understand fully the trade-offs one
undertakes when adopting a library such that (a) in most cases one
indeed does not want to adopt a library but (b) in exceptional cases
(such as qooxdoo and jsMath) the added value of the library hugely
exceeds the effort that will be needed to get along with them.

As a corollary, you seem unaware that I am Mr. NIH, not as a bug, but as
a feature. And I am using qooxdoo and jsMath ecstatically over how fast
I have brought a wysiwyg math editor to the web.

Is your brain nothing but one great honking anti-library neuron?

kt

 

[David Mark]

//teamalgebra.com/[/url]
Apparently, I have a flaky keyboard...

That's as maybe, but the culprit here is likely Kenny's flaky app.

When I type "g=mc2==" on the [unbookmarkable tab],

That I don't mind as I don't think tabbed interfaces should mimic
navigation.  They should persist their state though (e.g. with
cookies, local storage, etc.)

Not unsurprising and it will require debugging a meg of dubious JS to
track them down.

Let's see what the browser/engine is first.

I already told you.

Team qooxdoo seems to have
run up the white flag on Opera key events.

They can't even make it work *with* browser sniffing? Some team
you've got there.

My investors (me) are
prepared to lose that market.

Odd for European developers to give up on a browser that is very
popular in Europe. Lately it has gotten a boost from MS offering it
as an IE alternative.

Of course if Mr. Mark can provide a
/universal/ HTML API for key events I'll just load his code and embed my
client code that uses it in a qx.ui.embed.Html widget and be singing his
praises too.

Do you read this group or just post? I've long since tamed the
keyboard (and yes, that includes Opera). But I don't write add-ons
for qooxdoo.

[...]

I waded thru more than a little qooxdoo code to determine they had left
the model property off the RadioGroupBox widget.

That's too bad.

Doing so was not the
end of the world. Sorry. k

Nothing is the end of world, is it? But one problem after another
indicates you would have been better off without qooxdoo.

 

[David Mark]

Nisse Engström wrote:
//teamalgebra.com/[/url]
Apparently, I have a flaky keyboard...
Or unsupported browser or both, but...

How do you quantify what you "support" with a browser sniffing
framework like qooxdoo?  Do you just take their word for it?

You lost me there. Are you saying their browser-sniffing can guess
wrong? On what percent of users? Hint.

And I believe Nisse is using the latest Opera.

Opera? A tragedy, yes? From qooxdoo:

No. It's an excellent browser, despite a few quirks.

"Opera issues
Opera is the most broken browser concerning keyboard events.

What does that mean? It can't handle typing "g=mc2=="? I assure you
it can.

It supports
enough to use the keyboard for navigation but complex key commands
cannot be detected.

That's a design mistake on the part of the qooxdoo developers and they
are attempting to deflect the blame to Opera.

All key combinations containing the Alt key cannot be detected.

Rubbish. And why would they attempt to do something so silly?

The modifier flags are only set in keypress events.

Also rubbish. They are making up wild stories. And what if it were
true?

This makes
qx.client.Command in Opera unable to catch most interesting keyboard
shortcuts.

You don't have the slightest idea what you are talking about (no
matter what your definition of "interesting").

Opera sets only the keyCode property of the DOM event.

As with the previous nonsense, that's completely and demonstrably
*false*. In other words, they don't know what they are talking about
either.

Because of that,
it is not possible to decide if the code in keypress events is a key
code or a char code.

Also pure fantasy. You really don't read this group at all, do you?
Bad form.

Since both codes may overlap, collisions may occur.
Nope.

In need for a consistent solution, the current key handler
implementation always prefers the special keys over the normal keys.

Technobabble intended to misdirect inexperienced (and gullible)
developers.

For
example, “s” and “F5” both have a code of 115 in Opera. In this case
qooxdoo would prefer “F5” over the “s” key."

LOL. You have bet on the proverbial bad pony.

Well, it might not be live yet, but I have code in the dev version now
that refuses to handle anything but webkit, gecko, and ie.

I highly doubt that, considering how many variations of Gecko and
Webkit exist. But, even if it were true, it's still a failure.

I'll keep plugging on this issue, but what happened was I tossed the
prevent handling at a low level then later tossed all handling of what
qooxdoo calls keypress.

I suggest you toss the whole thing.

The former was OK, the latter should have been
to still do the preventdefault but no more (ie. no round-trip to server
which was ignoring those events anyway).

You can't do round-trips on every key event. As you've been told
repeatedly, it's pure lunacy (and doomed to fail).

Are you hoping it is seventy, or hoping I get it sorted out tonight?

I really couldn't care less. Have I not mentioned that? I only
comment on your serialized career suicide note to caution others.

One thing you do not seem to be coming up to speed on as quickly as I
thought you might is that I understand fully the trade-offs one
undertakes when adopting a library such that (a) in most cases one
indeed does not want to adopt a library but (b) in exceptional cases
(such as qooxdoo and jsMath) the added value of the library hugely
exceeds the effort that will be needed to get along with them.

Considering how confused the qooxdoo developers are about basic
keyboard handling, it's hard to see their wares as anything
approaching exceptional.

As a corollary, you seem unaware that I am Mr. NIH, not as a bug, but as
a feature.
Translation?

And I am using qooxdoo and jsMath ecstatically over how fast
I have brought a wysiwyg math editor to the web.

You are hopelessly deluded.

Is your brain nothing but one great honking anti-library neuron?

We've been over that. Even a child could understand that eschewing
obviously bad libraries does not indicate any sort of general "anti-
library" sentiment.

 

[Kenneth Tilton]

Do you read this group or just post?

Just post. This NG is not really up to snuff, and deals mostly with
stuff I let qooxdoo worry about.

I've long since tamed the
keyboard (and yes, that includes Opera).

Cool, I'll check out your code.

But I don't write add-ons
for qooxdoo.

You might recall qooxdoo allows one to embed raw Html/js.

That's too bad.


Nothing is the end of world, is it? But one problem after another
indicates you would have been better off without qooxdoo.

No, "one problem after another" was jQuery, Dojo, and YUI and the code
was impossible to follow. qooxdoo actually started off with a problem
(over busy datagrid (does My Library have a datagrid?)) which was
insanely easy to fix and has only occasionally had issues while bringing
a rather hefty desktop application to the Web:

http://teamalgebra.com/

kt

 

[Kenneth Tilton]

What does that mean? It can't handle typing "g=mc2=="? I assure you
it can.

The question is what key event application is made available to the
client in the general case. I'll check it out now that I have the thing
somewhat gracefully coping whether or not TeX fonts are installed.

You can't do round-trips on every key event. As you've been told
repeatedly, it's pure lunacy (and doomed to fail).

....and working:

http://teamalgebra.com/

Go figure. By the way, I have not even made an effort yet to optimize
the round-trips, the editor is so responsive.

You seem unable to absorb the fact that this is an Algebra tutoring
platform, not a type-touching trainer. But I am happy to remind you as
needed.

Translation?

Generally you are right: libraries are more trouble than they are worth.
qooxdoo and jsMath are clear exceptions.

You are hopelessly deluded.


We've been over that. Even a child could understand that eschewing
obviously bad libraries does not indicate any sort of general "anti-
library" sentiment.

I guess not since you are pushing a competitor library /and/ selling
your services as a qooxdoo developer.

kt

 

[Kenneth Tilton]

//teamalgebra.com/[/url]
Apparently, I have a flaky keyboard...
That's as maybe, but the culprit here is likely Kenny's flaky app.
When I type "g=mc2==" on the [unbookmarkable tab],
That I don't mind as I don't think tabbed interfaces should mimic
navigation. They should persist their state though (e.g. with
cookies, local storage, etc.)
I get
"G0MC200" (and the "2" also triggers a browser shortcut).
And there are heaps of other peculiarities on that site.
Not unsurprising and it will require debugging a meg of dubious JS to
track them down.

Let's see what the browser/engine is first.

I already told you.

Team qooxdoo seems to have
run up the white flag on Opera key events.

They can't even make it work *with* browser sniffing? Some team
you've got there.

My investors (me) are
prepared to lose that market.

Odd for European developers to give up on a browser that is very
popular in Europe. Lately it has gotten a boost from MS offering it
as an IE alternative.

Yeah, it's going through the roof:

http://www.w3schools.com/browsers/browsers_stats.asp

2.1% and dropping from a high of 2.4 in December, 2008. Be still my
beating heart. Glad I checked before seeing if I could make it work.

kt