In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] (John Delacour)
wrote:
> I have exactly the same set-up. The only difference seems to be my
> pain threshold :-)
Yes, that whole extra second Mac::Glue takes over osascript can be annoying.
I am only being half-facetious, I know it can be, depending on the context.
But in most cases, it won't be significant.
You may not be aware, but unlike RunAppleScript(), the worst overhead
involved with Mac::Glue is on script startup only, whereas with
RunAppleScript() there will be slowness overhead associated with every call
to the function.
[As a side note: again noting that the delay is entirely in the startup, I
use Mac::Glue a lot in XChat scripts, and they are loaded in once and cached
a la mod_perl, so there is no startup issue. You get immediate results,
just as one would expect from a compiled AppleScript.]
> The problem with your routine is that you need to display the dialog
> in a process which a) has glue available and b) the user does not
> want to activate.
I don't understand the complaint. It's just like yours. Yes, you need a
glue, but that's a one-time operation ("% glue /Applications/Terminal");
it's not a big deal. And yours targets the application too; what's the
difference?
> Besides that, the user is required to learn a new way of expressing
> the terminology rather than simply using the familiar AppleScript way.
You are assuming people on this list are familiar with AppleScript. I make
no such (false) assumption. :)
> The routine using Mac::AppleScript seems so much simpler to me
Yes, of this I have no doubt.
> and it is possible to display dialogs without leaving the current
> application, whether it be Terminal or the editor (at least the one
> I'm using).
OK, I think you misunderstood. It works just fine using any app as the
target app, as long as it is scriptable, just like yours.
> done no precise timings. Neither method is fast and 3 seconds to me
> is an awful long time if you think what that might have taken on a 66
> MHz machine.
Three seconds *including clicking Cancel*. So it relies on silly user
interaction, hence the note of the user/sys CPU times. Here's a quick
similarly "dumb" benchmark that relies on user interaction:
[EMAIL PROTECTED] pudge]$ time osascript -e 'tell app "Terminal" to choose file
name with prompt "test1" default name "test2"'
real 0m1.136s
user 0m0.200s
sys 0m0.070s
[EMAIL PROTECTED] pudge]$ time perl -MMac::Glue -e 'my $glue = new Mac::Glue
"Terminal"; $glue->choose_file_name(with_prompt => "test1", default_name =>
"test2");'
real 0m2.398s
user 0m1.120s
sys 0m0.210s
Woohoo, I clicked "Cancel" faster on that run. :) The real number is 0.27
vs. 1.33 there. About a second difference.
> use Mac::AppleScript qw(RunAppleScript);
> $asresult = RunAppleScript <<AS;
> tell app "Terminal"
> display dialog "Enter a number:" default answer "55"
> text returned of result as real
> end
> AS
> print $asresult**2 . $/;
use Mac::Glue;
my $glue = new Mac::Glue 'Terminal';
my $result = $glue->display_dialog('Enter a number:', default_answer => 55);
print $result->{text_returned}**2, "\n";
--
Chris Nandor [EMAIL PROTECTED] http://pudge.net/
Open Source Development Network [EMAIL PROTECTED] http://osdn.com/
