developer, geek, lover of cute stuff

Timisoara, Romania
Joined May 2007
Cristian Vat 🦄 retweeted
php-rust: Francesco Tinti’s Rust rewrite of PHP 8.5 now runs full WordPress core tests at near-parity on real MySQL.
48
96
50
1,136
80,468
I know I am 4-6 months behind on AI usage...and yet I see people who seem to be much farther behind...
28
Cristian Vat 🦄 retweeted
Everything got pwn'd. I tried to warn you.
154
442
44
5,273
336,771
Cristian Vat 🦄 retweeted
Replying to @hd_nvim
the actual valid PEM using curve P-384 would look like something like below, but not fun -----BEGIN EC PRIVATE KEY----- MIGkAgEBBDAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAASgBwYFK4EEACKhZANiAAQTglHNUqySmMHIqtl3Mh3rl+cJvQtMoKyl XcitUdz8nRWJoVl+OlEg4e/WMcY+GDXKyuKYaaYuFjHoooGBq1ZhbcRdkYq8CfOr DmPPeSqk3O1zh743u6VpVJ8cArJw7Wc= -----END EC PRIVATE KEY-----
5
193
4
2,839
38,071
didn't anyone rewrite PHP in Rust yet?
1
1
119
Cristian Vat 🦄 retweeted
Today we disclose PHP2Shell - remotely exploitable memory corruptions in PHP WordPress pre-auth RCE: ✅ Drupal pre-auth RCE: ✅ Joomla pre-auth RCE: ✅ Remote memory disclosure of any PHP application: ✅ Read more ⬇️🧵
66
568
94
3,103
468,997
Cristian Vat 🦄 retweeted
I always thought a guest WiFi network was dumb. After cracking into my tv like a walnut with nothing but WiFi I no longer think this You’ll get the 2.5ghz channel and you’ll like it
Having Claude build me a smart home Mac tool and my biggest takeaway is to never give anyone WiFi access to my home ever again
24
119
10
6,162
233,393
Cristian Vat 🦄 retweeted
"Co-Authored-By: Claude Opus 5.5" is giving me far too much credit.
95
391
25
9,505
206,777
Cristian Vat 🦄 retweeted
did you use an agent or was it a hand job?
116
102
20
3,192
148,392
first spooktober project. detecting Crypt Keeper laugh and pulsing pumpkin light. Custom Tessel ( github.com/PatrickSt1991/tes… ) player build deployed on Samsung TV reporting video and play times to service which has subtitles and controls IKEA light tied to DIRIGERA hub
1
1
141
and it works, barring very minor timing differences in subtitles...
1
40
the most complicated part was getting proper subtitles that have audio descriptions like "HA HA HA HA HA HA!" or "[LAUGHING]", opensubtitles had them... tried using several speech to text models but they don't output that at all it seems? they only give real 'speech'..
4
lazy human: I'll just let ChatGPT rename these .srt files. over-achiever Sol: renamed files, ran vlc probe to check, checked encoding of subtitle files, compared video/audio codecs with what's supported by my tv, compared timestamps of last subtitle entry with length of video..
1
124
just had some app ideas for my Fire TV stick but they also implied tailscale and...wtf why?!
1
98
@Tailscale is this region locked or something? ...
43
Cristian Vat 🦄 retweeted
We are thrilled to share that we are in development on a new animated series for Diablo with Netflix.
1,268
4,232
1,095
38,411
1,653,457
Cristian Vat 🦄 retweeted
Why use many token when few token do trick
3
3
36
3,058
yeah...this new era... hearing @dhh mention omacut on @lexfridman podcast, "hey this seems nice", then "oh boo it's linux-only", well...not anymore, you can just do things (while enjoying late lunch at a trattoria on a mac and remoting into a win machine for testing) 😀 github.com/deathy/omacut-win…
1
89
Cristian Vat 🦄 retweeted
31
187
20
1,760
39,289
I get it that commit message co-author started as marketing from Anthropic but I actually like it and enforce it for other models as well. maybe librarian mindset but N yeara in the future if some research says Opus 4.X always messed up validation of certain types of input I can give the newer models exact commits to check. Also I can see commit from Sol and know to ask Opus for review and vice-versa. It's not a bad thing, it's useful information and if you don't add it now you will not have it later. And you can't predict all future use-cases.
1
3
181
I mean in IT if someone comes at an interview now and they say they will not use AI generally (without any arguments or just moral arguments, see Linus's opinion on the topic) you just show them the door. quite different in other fields for now.
1
96
I already had an established perspective on colleagues before like colleague A prefers arrays over lists, colleague B over-abtracts. seeing a commit from colleague A but with model X changes my review mindset. Not necessarily better since the model might have other peculiarities than the person but good information to have.
22