On May 17, 6:50 pm, Thomas 'PointedEars' Lahn <PointedE...@web.de> wrote:
> Xu, Qian wrote: > > Is it possible to detect whether a (confirmation or prompt) dialog is > > present?
> No, such dialogs are always modal. Script execution in the window from > which these have been issued is halted until they have been closed.
A dangerous mistake I once had big troubles with. In Gecko-based browsers including Firefox build-in dialogs are modal only withing the current execution context. Because timed calls creating new contexts, they are excluded from the lock. I have no idea why such "pseudo- threading" was allowed for external scripts. Either indeed some reasons for that or some initial bug which is too late to fix. No any other known to me engine does that. For instance "enjoy" three alerts displayed at once on Firefox:
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html lang="en_US"> <head> <meta http-equiv="Content-type" content="text/html; charset=iso-8859-1"> <title>Demo</title> <style type="text/css"> * { box-sizing: border-box; -moz-box-sizing: border-box; -ms-box-sizing: border-box;
}
p { font: 1em Georgia, serif;
}
</style> <script type="text/javascript">
var counter = 0;
function demo() { window.alert('Pseudo-thread #' + ++counter);
}
function init() { window.setTimeout('demo()',500); window.setTimeout('demo()',1000); window.setTimeout('demo()',1500);
"Xu, Qian" <quian...@stud.tu-ilmenau.de> writes: > Is it possible to detect whether a (confirmation or prompt) dialog is > present?
No. While the dialog is being presented, script execution on the page is stopped. So, from another perspective, it is trivial. If your code is running, no dialog is being shown.
> And furthermore, how to press OK, Cancel buttons of a dialog?
User interaction.
> I am writing a javascript based test framework. I would like to detect > and close dialog using javascript.
The traditional way that test-frameworks do this is to overwrite the alert and confirm functions with code that simulates a pressed button. The code using them doesn't need to know, it just gets a preconfigured answer back.
VK wrote: > On May 17, 6:50 pm, Thomas 'PointedEars' Lahn <PointedE...@web.de> wrote: >> Xu, Qian wrote: >>> Is it possible to detect whether a (confirmation or prompt) dialog is >>> present? >> No, such dialogs are always modal. Script execution in the window from >> which these have been issued is halted until they have been closed.
> A dangerous mistake
My wording was just imprecise.
> I once had big troubles with.
No offense meant, but given the nature and number of misconceptions you have demonstrated before, one wonders how you are able to write working code at all.
> In Gecko-based browsers including Firefox build-in dialogs are modal only > withing the current execution context.
I was aware of that.
> [example code]
Your examples could be much shorter if you omitted all the unnecessary parts.
PointedEars -- Use any version of Microsoft Frontpage to create your site. (This won't prevent people from viewing your source, but no one will want to steal it.) -- from <http://www.vortex-webdesign.com/help/hidesource.htm>
> > For instance "enjoy" three alerts > > displayed at once on Firefox:
> I've seen 3 successive alerts ... not at once :-)
I mean if you don't close each alert right away then in ~1.5sec you'll have three alerts at once on the screen: drag one to see other underneath etc. You may arrange them nicely on the screen, make a screen shot of it and send to Mozilla like "Who may guess what a hey is that?" :-) Maybe someone knows it. :-)
Nothing relevant to the script. I just grabbed the first template from the list to make the demo. Normally we are using it in combination with HTML Transitional doctype to have the normal box model for all supported browsers (IE6/IE7, Firefox etc.) See if interested http://www.quirksmode.org/css/box.html
>>> For instance "enjoy" three alerts >>> displayed at once on Firefox: >> I've seen 3 successive alerts ... not at once :-)
> I mean if you don't close each alert right away then in ~1.5sec you'll > have three alerts at once on the screen: drag one to see other > underneath etc.
No, I'm sorry but I haven't that (on my Mac) with Firefox. The box(es) is(are) centered and attached at the window title bar. With my old IE, Opera and Safari, dialog boxes are floating and ... no ... only one alert after the other.
>>> * { >>> box-sizing: border-box; >>> -moz-box-sizing: border-box; >>> -ms-box-sizing: border-box; >>> } >> what these styles are supposed to do ?
> Nothing relevant to the script.
Ha!? OK
> Normally we are using it in combination > with HTML Transitional doctype to have the normal box model for all > supported browsers (IE6/IE7, Firefox etc.) See if interested > http://www.quirksmode.org/css/box.html
That force IE to see borders as all other browsers ?
On May 18, 6:09 am, SAM <stephanemoriaux.NoAd...@wanadoo.fr.invalid> wrote:
> VK a écrit :
> >>> For instance "enjoy" three alerts > >>> displayed at once on Firefox: > >> I've seen 3 successive alerts ... not at once :-)
> > I mean if you don't close each alert right away then in ~1.5sec you'll > > have three alerts at once on the screen: drag one to see other > > underneath etc.
> No, I'm sorry but I haven't that (on my Mac) with Firefox. > The box(es) is(are) centered and attached at the window title bar. > With my old IE, Opera and Safari, dialog boxes are floating and ... no > ... only one alert after the other.
Strange. Is Firefox using a different script engine for MacOS? I'm going to check. For Windows and for Linux it is the same as I described. Here for instance Firefox under Mandriva Linux 2008.1 Spring: http://www.geocities.com/schools_ring/tmp/alert.png
> >>> * { > >>> box-sizing: border-box; > >>> -moz-box-sizing: border-box; > >>> -ms-box-sizing: border-box; > >>> } > >> what these styles are supposed to do ?
> > Nothing relevant to the script.
> Ha!? OK
> > Normally we are using it in combination > > with HTML Transitional doctype to have the normal box model for all > > supported browsers (IE6/IE7, Firefox etc.) See if interested > >http://www.quirksmode.org/css/box.html
> That force IE to see borders as all other browsers ?
It forces others plus IE to calculate the container size properly instead of W3C's "content only" schema. On big projects it saves an ocean of time and efforts. From the other side I already explained that I just grabbed the first template I had to make the demo. I didn't want nor willing to start rwars over box models. I personally do agree with Peter-Paul Koch and I do consider W3C's box model - after being tried many times - as hugely inconvenient for any practical use. Yet I am not pushing my opinion onto community - and definitely not at c.l.j.
> On May 18, 6:09 am, SAM <stephanemoriaux.NoAd...@wanadoo.fr.invalid> > wrote: >> VK a écrit :
>>>>> For instance "enjoy" three alerts >>>>> displayed at once on Firefox: >>>> I've seen 3 successive alerts ... not at once :-) >>> I mean if you don't close each alert right away then in ~1.5sec you'll >>> have three alerts at once on the screen: drag one to see other >>> underneath etc. >> No, I'm sorry but I haven't that (on my Mac) with Firefox. >> The box(es) is(are) centered and attached at the window title bar. >> With my old IE, Opera and Safari, dialog boxes are floating and ... no >> ... only one alert after the other.
> Strange. Is Firefox using a different script engine for MacOS?
I've always seen JS alert stopping function(s). Since NC3, at least on Mac. While a modal box is opened nothing in JS client side can happen. I think that if a setTimeout was launched before the alert, the time of wait continues to be counted whereas the box is opened. With your example if we close a box the following one opens at once. ( The waiting time is finished)
> The traditional way that test-frameworks do this is to overwrite the > alert and confirm functions with code that simulates a pressed button. > The code using them doesn't need to know, it just gets a preconfigured > answer back.
