Gina Peter Banyard

PHP core developer and technical writer for the PHP documentation, funded by @thephpf
Lead maintainer, and master of typos, for the French translation of the PHP documentation.
Conference speaker who's always happy to rant on stage.
BSc in pure mathematics from Imperial College London
Photograph/Otaku in my spare time
Speaks 🇬🇧🇫🇷🇩🇪🇯🇵

Public Key

npub1rz3q6msmelwyh0r2qczh0gga95l3uvrhdejucafxn0awvhextxyqtlv5cr

NIP-05 Address

Girgias@phpc-social.mostr.pub

Profile Code

nprofile1qqsp3gsddcdulhzth34qvpth5ywj60c7xpmkuewvw5nfh7hxtun9nzqpz4mhxue69uhhyetvv9ujumt0wd68ytnsw43qz9nhwden5te0wfjkccte9ehx7um5wghxyctwvsjezswc

Publishing to

relay.mostr.pub

relay.nostr.band

Author Public Key

npub1rz3q6msmelwyh0r2qczh0gga95l3uvrhdejucafxn0awvhextxyqtlv5cr

Show more details

Last Notes

2025-06-09 14:56:54

- reply

@nprofile…utw6 @nprofile…gd4r and null being rejected by scalar types has been a thing since PHP 7 with the introduction of them, so that's hardly anything new.

2025-06-09 14:54:13

- reply

@nprofile…utw6 @nprofile…gd4r this not new. PHP has multiple type juggling contexts.
You are trying to find consistency between the "function" type juggling context, which is one of the most strictest (and is the only one affected by the strict_types declare) and the comparison type juggling context, which is one of the most non-sensical one as it will coerce objects to int and emit notices.
I have a draft RFC looking into comparisons but it's all over the place & finding reasonable semantics is hard

2025-06-04 12:13:44

- reply

@nprofile…utw6 I mean, that's always the easy way out. A preprocessor for types could have existed all the way back in the PHP 7 era, and some have existed that long.
But I think it has mostly been that PHP has integrated a lot of stuff, so people don't feel strongly about it.

2025-05-31 11:50:11

Holy shit, during a discussion in a Discord with some friends the following image was dropped.
Yes this vote is from 2021, but holy shit both Israel and the US voting against this does reveal a bit of the game.
https://media.phpc.social/media_attachments/files/114/602/275/147/807/651/original/76b419dd52241b50.jpg

2025-05-27 13:59:27

Finally got around to publish my slides for my talk "PHP’s Type System Dissected" at @nprofile…2cge on my website.
https://gpb.moe/doc/slides/PHP-Type-System-Dissected-2025-04-DDD
You can also now find the recording of it on YouTube: https://www.youtube.com/watch?v=85fgBcV3BuM
And made public my repo for my talk "PHP: Pretty Horrible Programs" given at @nprofile…6zpj and @nprofile…8mve:
https://github.com/Girgias/pretty-horrible-programs
And if you lose this post, don't worry, you can find all of it on my website: https://gpb.moe/about !

2025-05-11 15:45:31

I do love being told that my attitude is the problem for creating an open source community by someone working on a technology with a foundation that has $250M+ in net assets that is relying on the technology I work on which has a foundation with less than $1M in net assets, who is bitching about how "badly" we implement something and that we should spend time scouring the internet to find the resource they maintain.
Instead of them reporting a bug report to our issue tracker.

2025-04-17 19:01:06

A #PHP meme idea was thought up today at @nprofile…2cge by Rouven #DDD25
https://media.phpc.social/media_attachments/files/114/354/835/963/783/377/original/ebfa5fd2a8c84e33.webp

2025-04-10 12:14:01

Oy people of #scotland and specifically #glasgow if you have nothing better to do on Saturday, how about a nice counterprotest?
https://bsky.app/profile/heatherherbert.uk/post/3lmbzto6jgc2l

2025-04-06 04:41:19

So I've been cooking this WE, and got a basic implementation of associated types for #PHP interfaces done.
https://github.com/php/php-src/pull/18260
https://media.phpc.social/media_attachments/files/114/289/179/228/736/108/original/7919b4b82c891847.png

2025-04-01 17:02:51

https://floss.social/@servo/114263266479139385 I don't have any stake in @nprofile…2lvc, but I very much agree with @nprofile…sl98 against the use of AI.
Already having need to deal with multiple people using AI to comment on #PHP doc issues to just regurgitate nonsense that confuses the author or said issue by trying to be "helpful" is a recurring thing that wastes my time and makes me angry.

2025-03-20 03:55:47

- reply

@nprofile…dhpg @nprofile…wjwc no PHP static analysis tool supports dependent types AFAIK (which a range type is, which is what is required to propagate this sort of information).

2025-03-18 15:56:01

- reply

@nprofile…pnag vous recherchez quel genre de sujet ?

2025-03-18 15:56:01

- reply

@nprofile…pnag vous recherchez quel genre de sujet ?

2025-03-13 13:42:47

- reply

@nprofile…gms0 yaaah, if you have "weekly incidents" (which also, how did they pull this metric out) then you are definitely not applying any lessons learnt or implemented any of the recommendations.
Which DOES happen in aviation, the amount of times the NTSB has said X accident could have been prevented if Y previous recommendations were implemented is far from non-zero.

2025-03-13 11:11:12

- reply

@nprofile…etl4 oh we are doing the firefighter strategy to raise money now? 😝

2025-02-20 00:15:21

- reply

@nprofile…240e @nprofile…8xz7 it was me :p which I'm fairly certain I have already said on list!

2025-02-10 17:23:28

- reply

@nprofile…dhpg @nprofile…rv5u it even worked in PHP 7 prior to the introduction of Stringable ;)
I voted in favour of adding Stringable at the time, but with some retrospective, it is just yet another side effect of strict_types being... Kinda bad

2025-02-10 17:15:51

- reply

@nprofile…dhpg @nprofile…rv5u it does in weak typing mode.
(Reason number 345627 why strict_types were a mistake)

2025-02-07 18:11:44

- reply

@nprofile…dhpg @nprofile…yupt @nprofile…tvs5 I mean that sounds like a neat idea, but I don't think we really track scope in any way. Semantically you can think of a `foreach` as attaching a closure defined within the `{}` if you add scope. Which might be the only way to do it without causing massive BC breaks.
@nprofile…75ud's work on block closure would be useful for that.

2025-02-07 18:01:43

- reply

@nprofile…dhpg @nprofile…yupt @nprofile…tvs5 The internals discussion is small but should shed enough light https://externals.io/message/115702

2025-01-11 18:07:36

- reply

@nprofile…dhpg @nprofile…d24a @nprofile…4pen oh they finally dropped 5.6 support?

2024-12-23 17:03:49

For people wanting to block the dumb Copilot prompt on the GitHub dashboard add `github.com##.copilotPreview__container` as a uBlock filter

2024-12-05 15:15:55

If anyone wants a freebie #PHP #RFC to do with a simple implementation, just add `array_zip()` which performs a "zip" operation.
PHP already has this feature, but it is hidden away in `array_map(null, $array1, $array2, ...)` which is... kinda stupid.
I'll endorse you on the ML, and I could do it myself, but there is so much stuff...

2024-11-30 12:54:54

"My business needs Kubernetes" is the technology equivalent of "My family needs an SUV"
#technology #joke

2024-11-21 09:53:54

- reply

@nprofile…utw6 @nprofile…dhpg @nprofile…x9zf @nprofile…8t8c considering people in the cryptographic community just think it is bad [1] I don't think it is *just* a people problem.
[1] https://soatok.blog/2024/11/15/what-to-use-instead-of-pgp/

2024-11-21 09:25:17

- reply

@nprofile…utw6 @nprofile…acfp @nprofile…x9zf yeah, and we did do a retag somewhat recently and it was hell and everything broke and we decided to never do this again.
Considering the only "valid" reason, IMHO, for the delay is now gone, let's just scrap it.

2024-11-21 09:23:51

- reply

@nprofile…dhpg @nprofile…utw6 @nprofile…x9zf I still don't understand the point of "signing" the release, it would frankly be better to build it in CI in a reproducible way and publish the build command+check sum of that.
(Also the more I hear about PGP the worse it gets seemingly)
Seems @nprofile…8t8c would be interested in that topic.

2024-11-20 17:42:41

- reply

@nprofile…utw6 @nprofile…x9zf I have *enough* stuff to do that fixing how PHP does releases is so far down the list I need not one but two clones of myself.
How releases are done is... just frankly broken, yet again a tag was messed up, so the first "proper" version of PHP 8.4 is going to be 8.4.1 because 8.4.0 does not include the security fixes.

2024-11-20 16:03:43

- reply

@nprofile…utw6 @nprofile…x9zf And this delay of 2 days is frankly dumb. These are security releases, and I don't know anyone that uses the binaries from the php.net website...

2024-10-19 22:52:04

- reply

@npub13w7…mktw @npub1n3u…cnm2 Not really, you can implement some monads just with generics (e.g. State, NonDeterminism) but if you want the Maybe/Option and Either/Result/Exception monads you need tagged unions.
If you *only* care about Monads for error handling, you can kinda get away with it with just tagged unions and without a generic bind operator.
I do wonder if having some sort of effect `function foo(): T!{error: E1}` to encode Midori's lightweight exception is a good idea.

2024-10-19 20:47:06

- reply

@npub13w7…mktw @npub1n3u…cnm2 arguably Monads also need some language support.
Objectively to have reasonable and correct implementation of a Monad you need generics and ADTs, and ideally a bind operator.

2024-10-19 17:58:08

- reply

@npub1n3u…cnm2 sure but that's just a Monad, and they have their own usability problems.
Effect handlers and effect typing solve these problems in a way nicer way. Just no mainstream PL implements them.

2024-08-09 09:35:17

- reply

@npub1jpp…u59l Similar to others, depends on the CfP and if one of my talks is accepted. :D

2024-08-08 14:22:09

I created 2 new issues on the php-tasks repo for people that want to contribute to #PHP:
- https://github.com/php/php-tasks/issues/31
- https://github.com/php/php-tasks/issues/32
Those issues are mainly about reviewing current code and see if it could be improved by using better, or more consistent engine APIs.

2024-04-05 07:55:31

Came back to Yoshino mountain after a week of good weather, and the difference in how many cherry trees have blossomed is astonishing.
However this is a #Mosstodon post!
Courtesy of Yoshimizu-jinja shrine on Yoshino mountain in Nara prefecture!
https://media.phpc.social/media_attachments/files/112/217/537/021/684/427/original/f640c54d0734abf7.jpg

2024-04-04 05:32:27

And some #Mosstodon photos I took today in Reikan-ji temple in east Kyoto!
https://media.phpc.social/media_attachments/files/112/211/311/995/944/521/original/fe7cfbfda35978aa.jpg
https://media.phpc.social/media_attachments/files/112/211/312/353/961/000/original/7a7736c7b0ae0782.jpg

2024-04-04 05:30:29

So #mosstodon photo I took yesterday under the rain in Nijō Castle in Kyoto
https://media.phpc.social/media_attachments/files/112/211/301/682/158/756/original/73006d1960ecf4c7.jpg

2024-03-25 13:38:56

I did take a rainy #Mosstodon photo in the #shukkeien garden tho!
https://media.phpc.social/media_attachments/files/112/156/602/205/344/740/original/75ecd01f795c2460.jpg

2024-02-17 03:22:18

- reply

@npub1jpp…u59l I could send you my graph theory notes which were really approachable in my opinion but it's way more theory than comp sci graph theory

2023-11-24 09:29:23

- reply

@npub1jpp…u59l @npub13nt…5ulg well yes.
Still doesn't change the fact that bugs after a refactoring might live on for a year (stares at SPL and some bug I introduced prior to PHP 8.2.0 being tagged)

2023-11-10 08:34:54

- reply

@npub1jpp…u59l Smh I think we did the same functions :p

2023-07-04 03:39:50

- reply

@npub1jpp…u59l The fact that after over a year and a half of @npub196k…ggck existence, contributing just over 10 000 USD will still put you in the top 20 financial contributors.
Related: everyone has opinions, but no one proposes solutions nor financing for said solutions.