always run finally blocks no matter how we leave the try block (RT #112099) #1
Conversation
|
I'll pass this around for people to eyeball today, and if there are no screaming objections, send it out as -TRIAL tonight. |
|
Cool, thanks. Note that this does change behavior a bit: |
|
I ... think ideally, if a user does that, a small gnome exits the monitor and hits the developer with a mallet. Also, that would quite possibly fix a bug in CGI scripts compiled to be persistent with an overriden exit. So, on the whole, LGTM |
| @@ -29,6 +29,8 @@ BEGIN { | |||
| *_HAS_SUBNAME = ($su || $sn) ? sub(){1} : sub(){0}; | |||
| } | |||
|
|
|||
| our @_finally_guards; | |||
There was a problem hiding this comment.
our @some_variable is generally perceived as an invite of please feel free to fondle me if you think it is justified. Even (or often especially) if rhe variable is _ prefixed.
In the case of T::T there is no scenario where such "fondling" is justified, therefore a my %stackstate with local() on a key is a better solution.
There was a problem hiding this comment.
I don't think that's going to be a problem in practice. Try::Tiny already contains the following fondleables: $VERSION, @EXPORT, _subname, _HAS_SUBNAME, and an entire hidden package: Try::Tiny::ScopeGuard.
karenetheridge
force-pushed
the
master
branch
from
a800324
to
838fb5e
Compare
|
Thanks! merged and released as 0.25-TRIAL. |
No description provided.