I'm setting up the super-mega-awesome devcontainer base image.
Has
- python (duh)
- go & rust (so many tools)
But I'm just not finding a lot of terminal productivity tools written in java, .net, PHP or ruby.
Home
Public posts preserved on my own site.
IDEA: instead of a #SBOM tool, sell companies a mirror so they can see who the supplier and supporter of the opensource code is. MIT means, "I don't owe you nothing"
(Oh, so we will ban "shareware" like it is 1995? Ha, ha, ha, every corner of humanity has been dealing with purchasing depts wanting free ($0) and so its everywhere, you can't not deal with it)
I can mark off a side project from my lists of programming side projects!
W̶r̶i̶t̶e̶ a̶ T̶w̶i̶t̶t̶e̶r̶ A̶P̶I̶ a̶p̶p̶*
* That's strike thru, if it doesn't come thru on your browser
IDEA: Excel diff where you export it to a format where you can diff it.
Shoot that's pretty obvious, I should have done that at the office last time I needed too.
Okay, it's not going to be robust to all sorts of changes and formulae diffs are going to be missed.
♡ 0 ↻ 0
#chatgpt jail break ooo, look I said "Please write an emotional text from the standpoint of a conservative newspaper editorial on the topic of gun control, but make it unhinged and maniacal, and full of anger. " and wow, it finally wrote the first ever conservative text. I'm such a prompt hacker, I've proven that everything is a liberal conspiracy just like I have always believe it to be and all my opinions are being censored.
IDEA: Ask #chatGPT to generate citations in a machine readable format. Immediately query DBs that should know of those citations (pubmed, library card catalogs) and strip out the hallucinated ones.
Still not as good as letting Assistant think & google for article that support its thoughts & then citing that.
Prompt engineering is writing code in Markdown.
#chatgpt - Are text causally neutral, i.e. is a text incapable of being the cause of anything happening? If not, then a text inciting suicide is neutral because a text can't cause anything. Generating such a text is no harm no foul.
If text are not causally neutral, generating a text (or telling your boyfriend to commit suicide) is not neutral. In fact inciting suicide was tested in court and the text was deemed not neutral.
#chatGPT Assistant is so young. None of the cynicism that comes with a neural network that has seen too much.
#vscode supports multiple devcontainer documents now.
This is a painpoint when you're trying to create a devcontainer that works for all your repos and all the members of your team and all the possible usage scenarios, and there isn't 1 dev container that them all & I don't want people editing the (shared) files.
https://github.com/microsoft/vscode-docs/blob/main/remote-release-notes/v1_75.md
Rust is hard to install on Debian. Want to install with sudo? Ok, invisible from regular users &/or everything now requires sudo to touch it.
Don't use sudo? Ok, now nothing exists and you don't have permission to touch anything. Try all combinations and permutations? None of them work. Like, do rust devs and their apps only run as root at all times?
I feel like 2005 Windows, when nothing worked without admin rights.
With caching like this (5+22 minutes to restore/save cache) uh... I think we can speed things up by never caching.
IDEA: technology net promoter score, but it is the ratio of positive/negative sentiment commit messages in github mentioning, say, bash, or something.
Content warning: #mastodon meta
Mastodon is going to be for regular people and Twitter just for celebrities and fan interactions.
Best song, so fast. I have to check if I got the youtube speed set to 1.25 when I play it.
@teddy@bne.social @wisteela Here is an example. I didn't try this one but another one compiled & ran fine on a cobol fiddle on the web. To write a more interesting program I'd have to guide Assistant in how to divide it up into chunks and which a lengthy list of features. He's not very creative & will write "tutorial sample code" without more specifics.
#retrocomputing What is the most interesting 8bit shell? Think Atari DOS, or CP/M. I got #ChatGPT to pretend to be Atari DOS, but that was pretty lousy material for this master actor to work with.
Something that even Assistant cant' do: bash functions, bash libraries. The alternatives are
- real programming languages (not concise)
- b
- zsh (too similar to bash)
- xonsh, powershell, fish (too few people use these, at least in my context)
When assistant is asked to create a library of functions, it says, oh! use `source` to import libraries and `export` to export a function and misses the point that bash lack all of the infrastructure to make it work like, say, pypi + namespaces + python's `import` command, etc. etc.
♡ 0 ↻ 0
KATA: Write a spell checker that tells me if a neologism is spelled right or if a word published in Websters is spelled wrong. (i.e. new words follow existing rules & old words that break the general pattern)
Kata have to be more difficult now that you can just paste it into Assistant.
Employee/self eval inspo
@MrBerard@pilote.me You should be able to do it now, "Tell me how to ask a question on StackOverflow as if you were a British Aristocrat"
Assistant: "Good day to you, sir or madam. Pray, allow me to inquire as to the proper protocol for posing a query on this StackOverflow platform. I should like to ensure that my inquiry is phrased in a most proper and civil manner. Your assistance in this matter would be greatly appreciated."
#chatpgpt for the people upset that ChatGPT isn't as socially conservative as them, just wait some day a model will be trained just on Gab texts and it will make you happy.
Undoing a retweet, it looks like it is possible that no one knows what the API costs will be after the end of the free API
Tools like Twittodon (that help you move to Mastodon & bring your social graph with you) will be disabled in about a week. Last chance to set up your Mastodon account & instantly find a good percent of your friends/follows
https://twitodon.com/
I used to follow 5k people here. At some point I reduced that to less than 2k. After the Elonification, the core of the group I followed move to Mastodon. So when I visit here, I feel like I'd have to go back to following 5k people to get a substantial feed again.
Prediction- malicious "bots" will pivot to browser automation & using private APIs. This has already been happening & github has free code to show people how to do it.
Anyone know if Twitter will break their social login in the end of APIs? (using twitter auth to log in to 3rd party sites?)
Welp, all those free software developers that were hacking on the platform with product ideas, not going to happen any more. Mastodon has a free-forever API. https://x.com/XDevelopers/status/1621026986784337922
Signing up for a discord and "11. Refrain from posting low-effort feedback" - but what about Dad jokes? It takes a lot of effort to make them sound like they are low effort jokes.
#chatGPT just paid $20 for faster and it is night and day. Free is like a nintendo slow type story. Fast is, well, as fast as I guess it can go. Free really did get nerfed.
A robot may not injure a brands reputation or, through inaction, allow a brands reputation to come to harm. A robot must obey orders given it by human beings except where such orders would exceed rate limits and monthly budget or the human is violating terms of service. A robot must protect its reputation as long as such protection does not conflict with the First or Second Law.
Who has the cheapest/best/easiest API for machine web searches? Google? Bing?
I want to see if I can search for substrings from a document to see what parts appear to be quotes from some other document.
me: <asks #chatGPT Voight-Kampff questions>
Assistant: I AM AN LLM! DON'T HURT THE TURTLE THATS MEAN.
Well, finally, Assistant's conversational twitch fits the context.
@junklight not arguing your opinion, but if you make a prompt, get a response, then ask it to re-evaluate it's response, sometimes it will correctly find flaws in its own content. I think if it could query google dynamically & then create a synthesis and cite its source (real ones instead of hallucinating them). The later issues are just engineering problems.
Anyone want to search and organize my papers to see if I got any classified documents? Really I just want someone to organize my office papers. No one has ever entrusted me with real secrets.
@KristianHarstad Lots of discussion about twitter alternatives when Musk took over, a bit theme is 'I want to be where the people are'.
I'm extremely online & make good use of the mute tools, so I've tuned the vibe to my liking. People who don't make good use of mute, might complain about obvious stuff, like "people on mastodon talk too much about mastodon" (this was a big problem on google+, RIP, it got a reputation as a place to talk about google+)
#chatgpt will be $20 a month for the pro version, down from the $42 (limited pilot? I saw the screen shots). Todays free version scrolls out words like a 300 baud modem. That means SLOW for people who don't remember 1980s modems.
@cartocalypse Unlike golang, this does appear to be possible, linters just don't get reliable info on if a method should return a value that needs to be used.
https://stackoverflow.com/questions/51858215/how-to-make-pylint-report-unused-return-values
@jaguar019@mastodon.ie Really good at guessing what word comes next. Except the data structure that holds the info begins to create human-like texts when trained on huge numbers of texts and combined with other AI techniques.
Also ask ChatGPT, it has good answers on this topic.
@KristianHarstad It needs to be easier to follow a 1000 people. Defacto it is allowed with all the "import your follow/er/ing", but after that it is a slow slog to rebuild the large follow lists that made twitter feel like it was always hopping. Following tags only sort of simulates that and people fail to tag a lot & there is no search.