Bob Weinand

Exploring every day the infinite possibilities of PHP and C

[Bob Weinand @bwoebi]

@hikari_no_yume Using firefox forever. But I fear, not enough people will switch and we'll just have sites completely broken in firefox.
Something needs to happen for people to leave Chrome in droves.

[Bob Weinand @bwoebi]

@hikari_no_yume (and yet, somehow the easily accessible large standard library of PHP is one of its biggest selling points. [Do you sometimes have to curse code so that people want to use it?])

[Bob Weinand @bwoebi]

@krakjoe Awesome. I still see that extension as a fundamental building block of some types of applications. Thanks for continuing to maintain it!

[Bob Weinand @bwoebi]

Put a new RFC under discussion: https://t.co/yTbNxhTrW3
Allowing default values while destructuring arrays.

[Bob Weinand @bwoebi]

@Ocramius With __debugInfo() creating a new object of itself. But given print_r is buffering, it'll just be memory exhaustion; use var_dump for the best effect.

[Bob Weinand @bwoebi]

@Ocramius git pull --rebase --autostash. Always. Almost always.

[Bob Weinand @bwoebi]

@Ocramius You mean an inspector like https://t.co/78OS1nfVig?

[Bob Weinand @bwoebi]

@hikari_no_yume No. But this just looks ugly. To torture your code reader, I recommend using random inconsistently bracketed initializer expressions. (I had to write code to parse these ...)

[Bob Weinand @bwoebi]

@Ocramius If you were masochistic, you could've written this using reflection instead.

[Bob Weinand @bwoebi]

@hikari_no_yume which you can disable in power settings (this fast startup thing)
I've seen all sorts of things going subtly wrong with this "hot shutdown" over time, until I just decided to disable it and have booting take a couple seconds longer...

[Bob Weinand @bwoebi]

@krakjoe Which daemons have you eaten, touching the arcane invocations presented by PHP 5.4 code? Daemons, begone!

[Bob Weinand @bwoebi]

@dan_in_tweets @jrf_nl @gmazzap Please note that *may* rather means this RFC does not decide upon it. There's no reason why dynamic properties should be removed at all. While I voted yes here, I definitely don't want the attribute removed. The RFC was very close and removing altogether will likely be rejected.

[Bob Weinand @bwoebi]

@frankdejonge I'd pin the library on a particular commit hash and call it a day as user. But yes, I do realize most people would be complaining in that case. I hate that we're fixing a people problem in a technical way, but there's seemingly no better alternative.

[Bob Weinand @bwoebi]

@doubaokun Before Fibers we had generators, these functions were just as "incompatible". Fibers are not an event loop. A fiber is a building block of an event loop. Without proper event loop is does not even make sense to talk about incompatibilities. Everything blocking is "incompatible".

[Bob Weinand @bwoebi]

@krakjoe The solution should be both: parallel for managing cpu usage and fibers for efficient I/O scheduling. You can do multiprocessing, but doing that for every CPU intensive task is an expensive choice and ugly to manage in userland (esp. error handling).

[Bob Weinand @bwoebi]

@krakjoe @kelunik @heiglandreas @SenseException @SimonSprankel @Fabian_ikono There are also limitations with callbacks - which are somewhat more popular in PHP than java - there's no way to express without generics something like function<E extends Exception>(callable() throws E $cb) throws E (you could blindly throw Exception, but that doesn't help ...)

[Aaron Piotrowski @_trowski]

Introducing fibers for #PHP: https://t.co/wScTjlSfdc

RFC for inclusion in core: https://t.co/9xrzBzkOFg

Amp v3 will use fibers to return proper results from functions instead of promises, reduce boilerplate, and make async programming more intuitive. https://t.co/UJaX7JKLsJ

[Bob Weinand @bwoebi]

@derickr The JIT is an absolute non reason to bump (for me), just a subsystem of opcache. But yes, it was the motivation for some; others who supported the move though was that it "enabled" deprecation cleanup.

[Bob Weinand @bwoebi]

@derickr The main driver of the major bump was that a lot of warnings/deprecations are now errors/removed functionality. Justified from that perspective ... But I would like if we were to move to a version.patch versioning scheme without distinction between minor and major.

[Bob Weinand @bwoebi]

@krakjoe I'll gladly use a debugger for C code. Mainly because it's non-trivial dumping all the context at once. In PHP it's usually just a few var_dump()s. Mostly 1-3 times dumping is sufficient. Esp. with conditional breaks (in loops), setting the debugger up is not easier than dumping.