Is it possible to use the debug library while using tk.rb?
It seems to work only as long as the Tk mainloop is not entered…
From: Wybo D. [email protected]
Subject: debugging using tk.rb
Date: Thu, 19 Oct 2006 20:06:23 +0900
Message-ID: [email protected]
Is it possible to use the debug library while using tk.rb?
It seems to work only as long as the Tk mainloop is not entered…
Maybe, the following cannot help you …
If it is the reason of the trouble that “Tk.mainloop” doesn’t
return while the root (main) window is alive, please try to start
“Tk.mainloop” in a Thread (e.g. evloop = Thread.new{Tk.mainloop}).
Hidetoshi NAGAI wrote:
return while the root (main) window is alive, please try to start
“Tk.mainloop” in a Thread (e.g. evloop = Thread.new{Tk.mainloop}).
when I do that literally, the script exits immediately, without any
messages. When I use () instead of {}: evloop = Thread.new(Tk.mainloop)
then it starts normally, but there is no difference with just
Tk.mainloop.
For example, running the Pig Latin Generator from Dave’s book
(attached): if I try to debug the pig method with
break PigBox:pig', then hitting thePig It’ button should stop me in
the pig method, but that doesn’t happen. (I can, however) circumvent
this by setting the break point on the line number).
And when I hit the `Exit’ button, an error message pops up:
FATAL
FATAL
while executing
“rb_out c00002”
invoked from within
“.w00000.w00004 invoke”
(“uplevel” body line 1)
invoked from within
“uplevel #0 [list $w invoke]”
(procedure “tk::ButtonUp” line 22)
invoked from within
“tk::ButtonUp .w00000.w00004”
(command bound to event)
One of my real, rather complex, applications says, as soon as I enter
the Tk-window with my mouse:
/home/wybo/bin/adm/admin:4:$: << File.dirname(FILE) files
(rdb:1) b ~/bin/admin_transaction.rb:108
Set breakpoint 1 at ~/bin/admin_transaction.rb:108
(rdb:1) c
/usr/local/lib/ruby/1.8/tk/event.rb:443: invalid value for Integer: "??"' (ArgumentError) from /usr/local/lib/ruby/1.8/tk.rb:1323:incatch’
from /usr/local/lib/ruby/1.8/tk.rb:1323:in callback' from /usr/local/lib/ruby/1.8/tk.rb:1557:inmainloop’
from /usr/local/lib/ruby/1.8/tk.rb:1557:in `mainloop’
from /home/wybo/bin/adm/admin:1248
/usr/local/lib/ruby/1.8/tk/event.rb:443: TkUtil.eval_cmd(cmd,
*(ex_args << klass.new(*klass.scan_args(keys, arg))))
(rdb:1)
I suspect that this error message is caused by the use of rio(??) (for
the creation of temporary files)
Hi –
On Sat, 21 Oct 2006, David V. wrote:
Wybo D. wrote:
When I use () instead of {}: evloop = Thread.new(Tk.mainloop)
then it starts normally, but there is no difference with just Tk.mainloop.Duh.
The discourse on this list seems to have drifted toward the impatient,
critical, and judgmental. Could it perhaps drift back toward
collegiality, based on the fact that we’re all involved in the ongoing
process of grasping and productively using various technologies?
David
[email protected] wrote:
Duh.
The discourse on this list seems to have drifted toward the impatient,
critical, and judgmental. Could it perhaps drift back toward
collegiality, based on the fact that we’re all involved in the ongoing
process of grasping and productively using various technologies?David
thanks, David Black, for your support.
The problem is, I’m one of those non-professional users who can, with
their limited knowledge, do a lot with Ruby without understanding all
it’s ins and outs. For instance, I have no experience with threads at
all, so when I am suggested a solution using those, I just try, and if
it works I have another reason to study threads in detail. Before that
I’m stupid enough to think that Thread.new{…}, which did nothing, may
have been a typo…
From: Wybo D. [email protected]
Subject: Re: debugging using tk.rb
Date: Sat, 21 Oct 2006 17:12:05 +0900
Message-ID: [email protected]
For example, running the Pig Latin Generator from Dave’s book
(attached): if I try to debug the pig method with
break PigBox:pig', then hitting thePig It’ button should stop me in
the pig method, but that doesn’t happen. (I can, however) circumvent
this by setting the break point on the line number).And when I hit the `Exit’ button, an error message pops up:
FATAL
I see. But unfortunately, there is no solution.
The callback process of Ruby/Tk is a little complex.
Longjmp must be handled very carefully, because it may break
consistency of Ruby’s stack, Tcl/Tk’s stack, and Tcl/Tk’s
internal status.
If not, such like as your trouble, longjmp causes SEGV easily.
Which version of Ruby/Tk do you use?
One of my real, rather complex, applications says, as soon as I enter
the Tk-window with my mouse:/home/wybo/bin/adm/admin:4:$: << File.dirname(FILE) files
(rdb:1) b ~/bin/admin_transaction.rb:108
Set breakpoint 1 at ~/bin/admin_transaction.rb:108
(rdb:1) c
/usr/local/lib/ruby/1.8/tk/event.rb:443: `invalid value for Integer:
“??”’ (ArgumentError)
That is fail of converting a event parameter to an object.
Tcl/Tk returns a string “??” for an undefined parameter on a event.
In that case, Tcl/Tk returns a “??” for the parameter which is
expected as an integer.
It may be a bug on Ruby/Tk.
Could you show me your definition of binding?
Probably, you will be able to avoid the trouble
by using “%” substiturion arguments.
Wybo D. wrote:
When I use () instead of {}: evloop = Thread.new(Tk.mainloop)
then it starts normally, but there is no difference with just Tk.mainloop.
Duh. Ruby is a strictly evaluated language, arguments to a function call
are evaluated before the call. Tk.mainloop hangs the interpreter before
Thread.new is ever called.
(You might want to spend some time on learning Ruby syntax and what the
difference between those brackets is.)
David V.