p5-Text-Xslate
p5-Text-Xslate copied to clipboard
wishlist: moving from Mouse to Moo
Are you planning to move from Mouse to Moo?
I have no plan to move to Moo.
Why do you recommend a slower module?
maybe use p5mop if it should ever become faster than Mouse :)
at least, most of the module is now written in xs, but still alpha software
https://github.com/stevan/p5-mop-redux
p5mop still has a long way to go before it begins to approach the speed of Mouse (the goal is to get there eventually, but that's still a ways off).
I don't hesitate to move to p5mop in a future, but for now it requires Perl 5.16.0 so I can't use it for Xslate.
I'd also prefer Moo over Mouse as the former is already in my (quite big) stack of CPAN modules and the later isn't.
While I personally prefer to use Moo most of the time, Mouse(::XS) is much faster and speed is most of the attraction for me in Xslate. So Moo-- here.
After an "interesting" week-end spent at work over the "fatal warnings" switch, I'd rather have Text::Xslate stick to Mouse.
If I stop using Mouse I prefer to use plain old Perl objects instead of any other OOP framework, but currently I have no time to make such a refactoring anyway.
I'm willing to review patches which drop dependencies on Mouse.
Have you considered Moo
with Class::XSAccessor
?
I think use of plain perl objects is preferable. Mouse dependency creates additional overhead in projects, where Moo or Moose used. As i see in sources - Mouse used only in few classes, so porting to perl object would be easy.
Use of plain perl objects looks the best solution that I, if any, should make. I hope someone make a pull request about it.
@gfx++
Also it maybe useful to switch to JSON::XS instead of Data::MessagePack, because JSON::XS is approximately 4 times faster.
As usual YMMV, but with a quick Benchmark script, I found DM to be ~70% faster than JSON::XS when serializing, and ~40% faster when deserializing. I'd rather have DM stay.
Is the plan to remove the dependency on Mouse still alive?