|
Absurd Alhazred posted:Trust. AHAHAHA. TRUST! OHOHOHOHO! TRUST! Reflections on 🤞-ing 🤞.
|
#
?
Jan 13, 2022 11:47
|
|
|
|
| # ? May 14, 2024 10:45 |
|
|
Debugging through some old C++ Win32 code that creates a PDF, using some kind of PDF driver and drawing stuff to a device context. If you stop the debugger at the wrong time, it just completely locks the machine up - freezes everything even down to the mouse pointer. I've had to restart my dev machine twice today already. (This work is with the aim of replacing the C++ code with something better written in .NET)
|
|
#
?
Jan 13, 2022 14:51
|
|
|
bolind posted:Absolutely, and same for Python. This is true of nearly every problem in computing though, and it was usually solved in the 70s
|
|
#
?
Jan 13, 2022 21:02
|
|
|
Ask me about the codebase I once worked on (initiated in the late 90s) that had no less than three different, in-house implementations of a doubly linked list.
|
|
#
?
Jan 13, 2022 21:45
|
|
|
bolind posted:Ask me about the codebase I once worked on (initiated in the late 90s) that had no less than three different, in-house implementations of a doubly linked list. All of them with their own unique memory leaks.
|
|
#
?
Jan 13, 2022 21:48
|
|
|
Volmarias posted:This is true of nearly every problem in computing though, and it was usually solved in the 70s ... in FORTRAN, and everybody's too scared to mess with it so it's only accessed through a library.
|
|
#
?
Jan 13, 2022 21:49
|
|
|
bolind posted:Ask me about the codebase I once worked on (initiated in the late 90s) that had no less than three different, in-house implementations of a doubly linked list. Any old enough codebase will have several different implementations of basic containers.
|
|
#
?
Jan 14, 2022 03:05
|
|
|
Zopotantor posted:Any old enough codebase will have several different implementations of basic containers. My favorite was the threading collections one of my old jobs had. Naturally they weren't thread safe. This didn't cause issues in production because all of its uses were accidentally bound to the main thread.
|
|
#
?
Jan 14, 2022 03:49
|
|
|
bolind posted:Absolutely, and same for Python. Fun fact: Python's unicode string type does not actually hold necessarily valid unicode. There's a hack added in python 3.1 where it maps bytes that aren't valid UTF-8 into recognizably broken surrogate pairs that let it reverse it back to the original bytes as long as you just hand the value back to one of its system outputs.
|
|
#
?
Jan 14, 2022 04:51
|
|
|
I like JSON because "use UTF-8 or else gently caress you" is a good call.Foxfire_ posted:I'm like 90% sure argparse on python doesn't technically work on POSIX in general because it assumes arguments are unicode strings instead of arbitrary binary. respect_for_python --; seems a feature that should not exist. if somebody really wanted or needed to read binary has input data, then have a mechanism where python receive the path of a file, and read from that file.
|
|
#
?
Jan 14, 2022 08:39
|
|
|
Tei posted:I like JSON because "use UTF-8 or else gently caress you" is a good call. You misunderstand. This is for file paths, not data. It exists because otherwise it would be impossible to represent arbitrary POSIX paths as Python strings.
|
|
#
?
Jan 14, 2022 12:21
|
|
|
Soricidus posted:You misunderstand. This is for file paths, not data. It exists because otherwise it would be impossible to represent arbitrary POSIX paths as Python strings. No, no,... my suggestion was to standarize over maybe UTF-8, and serialize text not compatible with UTF-8 has UTF-8. Like in JSON where you can have a text in Latin1, but is escaped so when is unescaped you get the original bits. Maybe is more complicated than that, and I am wrong.
|
|
#
?
Jan 14, 2022 12:40
|
|
|
Tei posted:No, no,... my suggestion was to standarize over maybe UTF-8, and serialize text not compatible with UTF-8 has UTF-8. So you have a valid unicode string, but you pass that to fopen() and it silently operates on some different text to what you passed? That seems like an even bigger footgun to me. The current scheme has the advantage that valid unicode doesn't get touched either way by the escaping.
|
|
#
?
Jan 14, 2022 13:52
|
|
|
Does Python not track encoding on a per-string basis and provide means to convert as appropriate? Ruby does that.
|
|
#
?
Jan 14, 2022 19:36
|
|
|
ExcessBLarg! posted:Does Python not track encoding on a per-string basis and provide means to convert as appropriate? Ruby does that.
|
|
#
?
Jan 14, 2022 22:11
|
|
|
recognize that unix paths are a land of insanity. respect those who want to make their own lives difficult by refusing to support dumb file names
|
|
#
?
Jan 15, 2022 05:24
|
|
|
Foxfire_ posted:Unix filenames are a dumpsterfire anyway and essentially nothing handles them correctly. They are arbitrary binary blobs, not text. Oh its worse than that. Back when I first learned Unix (pre linux) I worked out you could put VT-100 control sequences into file names. And the VT-100 control sequences where borderline turing complete. You could remap keys, tell the terminal to send out arbitary commands back to the mainframe, the whole lot. And all your victim needed to do was type "ls". I had a *lot* of fun screwing around with that in the computer lab at uni. I dont think Linux , or modern terminals, are quite that bad, but you can still do stupid poo poo with file names. Though the optional case insensitivity in macos can cause *very stupid* poo poo to occur.
|
|
#
?
Jan 21, 2022 04:17
|
|
|
You can still put control sequences in filenames, but common things like ls will escape them, and most modern terminal emulators refuse to implement control sequences they consider security risks, or at least disable them by default
|
|
#
?
Jan 21, 2022 10:53
|
|
|
Just has the mythical “can’t you just put everything in one table?” question from a client. The everything is 800+ fields for a “dynamic” form builder.
|
|
#
?
Jan 26, 2022 01:25
|
|
|
BigPaddy posted:Just has the mythical “can’t you just put everything in one table?” question from a client. The everything is 800+ fields for a “dynamic” form builder. Well if they're all atomic, independent, and linked only to the same primary key... it can technically still be up to fifth normal firm with just one table. Although if there's a case that screams for sixth form it's this one.
|
|
#
?
Jan 26, 2022 02:16
|
|
|
BigPaddy posted:Just has the mythical “can’t you just put everything in one table?” question from a client. The everything is 800+ fields for a “dynamic” form builder. No problem boss! *adds json column*
|
|
#
?
Jan 26, 2022 05:59
|
|
|
pokeyman posted:No problem boss! *adds json column* No lie, this is exactly what im doing for something instead of some fuckery with table inheritances in SQLAlchemy and Flask. I figure given that all reports will share the same basic header fields, all of the data (most of which is crunched behind the scenes) goes in a single TEXT field as JSON.
|
|
#
?
Jan 27, 2022 23:46
|
|
|
That is what is happening and now it is all  reporting!Ok well take the JSON and put it in your BI tool of choice and do all the reporting but I want it all in the same place as the app!We can show the reports made in the BI system in the app but I want this to be self serve without any developers needed to maintain it.
|
|
#
?
Jan 27, 2022 23:55
|
|
|
https://twitter.com/mjg59/status/1486462124897361922
|
|
#
?
Jan 28, 2022 04:50
|
|
|
if you click on a tweet now, and you are not logged in to a twitter account, if you scroll down a little bit to see replies you get a giant loving plaque that appears over the screen asking you to sign up or log in. and it does not have a close button twitter is cancer
|
|
#
?
Jan 28, 2022 12:07
|
|
|
Hammerite posted:twitter is cancer yes, but we knew that well before they started their persistent login popups
|
|
#
?
Jan 28, 2022 13:15
|
|
|
@mjg59 posted:1) It links the email address in a commit to a github user with that address no matter who committed it @mjg59 posted:2) If a commit is pushed to a fork, it's visible in every fork (including the original)
|
|
#
?
Jan 28, 2022 14:42
|
|
|
ExcessBLarg! posted:I never liked that Git prioritizes showing the Author of a commit over the Committer (except in --format=fuller). Both are really just advisory but the former even more so. ... Wouldn't the only committer in original Linux use of Git have been Linus, or would it be the first person to have accepted the patch into their tree that then got merged into Linus's? Anyway, either way it sort of makes sense for Linux-like set up.
|
|
#
?
Jan 28, 2022 14:55
|
|
|
Hammerite posted:if you click on a tweet now, and you are not logged in to a twitter account, if you scroll down a little bit to see replies you get a giant loving plaque that appears over the screen asking you to sign up or log in. and it does not have a close button Hasn't this been a thing for a few years?
|
|
#
?
Jan 28, 2022 15:32
|
|
|
Hammerite posted:if you click on a tweet now, and you are not logged in to a twitter account, if you scroll down a little bit to see replies you get a giant loving plaque that appears over the screen asking you to sign up or log in. and it does not have a close button If you clear your website data for twitter it goes away (for awhile) Or just stop reading twitter, probably the better answer
|
|
#
?
Jan 28, 2022 15:54
|
|
|
HappyHippo posted:Or just stop reading twitter, probably the better answer I probably would if people would stop posting tweets on this website, but that cat is out of the bag
|
|
#
?
Jan 28, 2022 16:03
|
|
|
If you block all twitter cookies entirely it stops showing up, if they're evercookieing you regardless then they've decided not to make that obvious.
|
|
#
?
Jan 28, 2022 16:23
|
|
|
Hammerite posted:I probably would if people would stop posting tweets on this website, but that cat is out of the bag I'll try and screencap/copy-paste text for tweets properly in future, sorry!
|
|
#
?
Jan 28, 2022 16:52
|
|
|
Absurd Alhazred posted:I'll try and screencap/copy-paste text for tweets properly in future, sorry! I was not criticising you for posting tweets. People post embedded tweets all over the forums, I hate it personally but there is nothing to be done about it, it's a done deal. The embedded tweet showed up just fine, I visited Twitter to see more context but got blocked by that stupid log in prompt.
|
|
#
?
Jan 28, 2022 17:42
|
|
|
Hammerite posted:if you click on a tweet now, and you are not logged in to a twitter account, if you scroll down a little bit to see replies you get a giant loving plaque that appears over the screen asking you to sign up or log in. and it does not have a close button Instagram started requiring login for viewing linked insta posts a good while ago makes some newspaper stories dumb to read Instagram, meet Ernst Germ (69)
|
|
#
?
Jan 28, 2022 17:58
|
|
|
Hammerite posted:I was not criticising you for posting tweets. People post embedded tweets all over the forums, I hate it personally but there is nothing to be done about it, it's a done deal. The embedded tweet showed up just fine, I visited Twitter to see more context but got blocked by that stupid log in prompt. In the past I would often post screencaps at least because I would post controversial tweets that I would worry would get deleted; I hadn't realized Twitter has gotten even worse on follow-through (they're bad enough for active users!). Seems like good practice to not have people be forced to depend on either embeds or follow-through for content and context. 
|
|
#
?
Jan 28, 2022 18:01
|
|
|
Hammerite posted:if you click on a tweet now, and you are not logged in to a twitter account, if you scroll down a little bit to see replies you get a giant loving plaque that appears over the screen asking you to sign up or log in. and it does not have a close button Nitter.
|
|
#
?
Jan 28, 2022 18:23
|
|
|
Absurd Alhazred posted:In the past I would often post screencaps at least because I would post controversial tweets that I would worry would get deleted; I hadn't realized Twitter has gotten even worse on follow-through (they're bad enough for active users!). Seems like good practice to not have people be forced to depend on either embeds or follow-through for content and context. There's always twitter proxies that strip all that stuff like Nitter, but that doesn't embed in the forums.
|
|
#
?
Jan 28, 2022 18:43
|
|
|
Nitter, plus a browser extension that redirects Twitter links to Nitter.
|
|
#
?
Jan 28, 2022 19:30
|
|
|
|
| # ? May 14, 2024 10:45 |
|
|
Passwords and whatnot changed for obvious reasons. Found this gem in an application provided by my company's offshore resources:  That's two sets of hardcoded credentials. In clientside Javascript. 
|
|
#
?
Feb 1, 2022 17:57
|
|
