The biggest problems I've had have been with new releases of Perl/Tk breaking old code. For instance, here's a patch I had to make because Tk804.025 broke my app:
2915c2924,2926
< sub{$self->clicked_on_assignment($self->{ASS_KEYS}->[$ass_
+lb->curselection()])}
---
> # sub{$self->clicked_on_assignment($self->{ASS_KEYS}->[$as
+s_lb->curselection()])}
> # ... used to work, but no longer does
> sub{$self->clicked_on_assignment($self->{ASS_KEYS}->[$name
+_to_key{$ass_lb->get($ass_lb->curselection())}])}
Of course, it's always possible that I was depending on behavior that was actually not supposed to be documented and reliable behavior. I'd have to go back and look at the docs I was using (the O'Reilly Mastering Perl/Tk book) and see. | [reply]
[d/l] |
Well, here's a recent example that's near and dear to my heart, I recently discovered that using the widgets as HList items, when one deletes an item using $HList->delete('all'), if the item has any widget entries, they will not be destroyed. Perhaps I am wrong in assuming that this should happen, but there is nothing in the documentation to lead me to believe that behavior should be otherwise. If I destroy a widget, it's children should be destroyed as well. I was under the impression that deleting an item (or an entry) was equivalent to destroying it.
With the Optionmenu widget, if you use an option list like [[label1=>1],[label2=>2]] you cannot simply set the -variable option, if you want the correct value displayed, you must also set the -textvariable option. If you only set the variable, the value in the variable will be overwritten by the first item in the option list.
These may be bugs, or they might be deficiencies in the docs. I don't care, I just don't like having to experiment to discover these behaviors.
| [reply]
[d/l] |
Tk804.027 delivered with Perl 5.8.7 doesn't render accented chars in Tk::Text properly. It's on windows, don't know if it's same on unix OS. So one has to downgrade Perl to older versions.
Generally no problem but for instance when you intend to use PAR to make an executable using Tk and DBD::CSV as in my case, I came across a bug which crashed the exe. Only the latest PAR version with Perl 5.8.7 worked without crashing but this combination renders garbled chars.
rahed | [reply] |
| [reply] |