> Also cc'd to
> tcltk@perl.org
> https://www.nntp.perl.org/group/perl.tcltk/2018/07/msg718.html
> per Vadim's request
Thanks
> I think it is important to discriminate between the end user of a tck/tk
> application and the programmer of that application. What would an end user
> think if he started getting flooded with warnings for a existing application
> they use that they cannot control, that only the application developer can
> modify.
Correct; the only good solution is to have correct implementation of refs
disposal;
BTW perl/Tk happen to have this problem solved; Tcl.pm having some difficulties
due to different approach on the problem.
>
> I too think that 100 is too small a threshold. I have a
> ....
> To tell you the
> truth, under v1.02 i did not see a performance penalty to introducing all
> these
> entries to $anon_refs and the tcl<=>perl command table. This leads me to
> believe that 100 may be way to small and even 1000 might be as well.
Any number will do! :)
BTW are you aware that, in terms of performance, this:
sub WIDGET_CLEANUP() {0}
if (WIDGET_CLEANUP) {
...
is much better than this:
our $SAVEALLCODES; # ...
...
if ($SAVEALLCODES) {
...
It's OK for development, but rather significant for release;
