kouiskas

A terminal manager for Claude Code: https://github.com/gi11es/deckard

The main goal is to reduce cognitive load managing many Claude Code and terminal sessions in parallel, while keeping things simple with a design focused on peripheral awareness of session status. And making everything resumable. The Claude session explorer also supports forking past sessions mid-conversation.

Overall this is designed to support the Claude-first software development workflows I've developed over the past year.

I’m talking about the library support, “libghostty”. It’s what cmux and others are built on top of. There’s nothing difficult about it, I’ve just found it very unstable in the way I used it. I’ve switched to SwiftTerm now.

Location: France

Remote: Yes

Willing to relocate: No

Linkedin: https://www.linkedin.com/in/gillesdubuc/

Email: gilles [at] dubuc.fr

Calendly: https://calendly.com/gi11es

Engineering leader with 20 years of experience on large scale web projects. I’ve remotely led fully distributed teams of engineers to architect, build, ship and run scalable distributed systems with up to 1B+ unique devices per month (Wikipedia). I’ve worked for large established organizations as well as fast growing startups. I love building healthy engineering cultures and growing engineers, helping them deliver their best work while working on complex projects.

I’m talking about concrete examples of well-known artists, let’s not confuse this with the delusions of sociopathic CEOs taking credit for other people’s work and constructing a fictional vision a posteriori for the media. There’s nothing artistic there.

On this debate people are confusing artistic illustration skills with creating art. An artist created this artwork at the state fair, not the AI. The artist had a vision, directed the AI towards that outcome and selected the result they wanted.

Some very famous artist exposed in museums today don’t physically hold the tools that « make » the artwork, they direct a team of assistants who act as their hands. Some like to hold the paintbrush themselves but they don’t have to and that’s why it’s not a criteria for whether they are the artist or not.

Much like a movie director directing a team to create a movie doesn’t hold the mic boom or deal with the photography themselves. They orchestrate other humans into creating the piece of art they have envisioned.

This is exactly the same. Instead of a known visual artist directing art students to fabricate the artwork they’ve envisioned, the artist directs the AI.

I charge my EV with free solar energy 90% of the time. Can't dig my own gas out of the ground for an ICE. Hopefully that kind of setup will become cheaper and more accessible to many.

EVs aside, every bare roof is a missed opportunity for solar anyway. It should be mandatory for every overground parking lot to be covered in solar panels, same for every newly built residential and office building. If this sort of policy was in place, there would be a lot more locally generated electricity for heating and charging EVs that doesn't need to be transported over long distances.

Wikimedia Foundation

Many positions are REMOTE (or in SF if you want), some are SF-only. VISA help/sponsorship provided for candidates willing to relocate to SF. There are also INTERN positions open right now.

http://wikimediafoundation.org/wiki/Work_with_us#Wikimedia_C...

It will happen later: https://www.mediawiki.org/wiki/Phabricator#Migration_timelin...

Some ideas...

- using tools like http://www.pictriev.com/ or similar offline solutions to get a pretty good estimate of her age.

- using image analysis to "fingerprint" the camera used to take the picture. Then crawling sites where she would be likely to post pictures (twitter, etc.) based on the suspected area and comparing to find matches (i.e. pictures taken with the same camera). I'm not sure what the current state-of-the-art is for those algorithms, but if for instance the camera displayed an obvious visible flaw (darker spot), it would have made it easier. On a perfectly clean camera without visible defects I assume that more than one picture is necessary for effective camera fingerprinting.

- using tools like http://graphics.cs.cmu.edu/projects/im2gps/ to try and determine the location just based on visual features.

They're lying or confused, the data I have is definite proof. 9gag's password and ours were hashed unsalted with MySQL's PASSWORD(). I'll reply on that comment.

They were using and probably are still using unsalted MySQL 4.1+ PASSWORD()

https://news.ycombinator.com/item?id=5677550

Also, in addition to confirming that our own hash was genuine, 9gag's password was very weak and their hash can be reversed with online rainbow tables.

name.com's passwords were (still are?) not encrypted. They're unsalted MySQL 4.1 PASSWORD() hashes.

The name.com sample data HTP showed in their log by querying the database was real. Source: I work for one of the companies they used as sample rows when querying the name.com DB. Our head of ops confirmed that the hash in the HTP log was indeed the MySQL 4.1 PASSWORD() unsalted hash of our password at the time. name.com is kind of generous with the term "encrypted" in their email.

Looking at the hash found in the query HTP ran, 9gag's name.com password was "harry1" at the time of the exploit. It also tells us name.com stores the passwords as unsalted MySQL 4.1 PASSWORD() hashes...

"sign language" should be labelled as ASL if that's what you meant. There are hundreds of sign languages.

Looks quite depressing, to be honest.

It's a photoshopped picture that's been around for years. On the original you can see how bad a copy/paste job it is: http://worldnewstweets.com/wp-content/uploads/2011/05/osama-...

Following that questionable logic, now that node.js is the latest hot fad, doesn't that make RoR the new PHP?

Average developers blame the language for their inability to make the best use of it. Good developers are too busy writing great code with the tools they have at hand to care.

Some languages have strengths for particular applications, but spaghetti can be written in anything.

And why should you care if 99% of the rest of the world writes shit code with the language you use, as long as your own house is clean?

In other words... promote synergy, direct workflow, like a boss.

Moving the monitor a bit further away from you and/or tilting it (which looks a bit unnatural, having your monitor "facing up" slightly) solves that problem for me.

I've only used 22" monitors vertically, maybe bigger ones are just too big for vertical use and worsen the issue you've experienced.

I had to spend a lot of time to find monitors that use technology with a decent vertical viewing angle.

Nowadays most LCDs use cheaper techs like TN panels that only give a good viewing angle when used horizontally. You have to look for monitors with IPS or S-PVA panels. The one you have has a TN panel, which is the worst kind for vertical use, I'm not surprised that you were disappointed.

Be careful and triple-check that the monitor you intend to buy uses the right panel technology, some manufacturers lie about it and others sell different techs for the same model number.

I have two Dell 2209WAB and one HP LP2275w (in different locations, I only use one monitor at a time + laptop). Both models provide excellent angles for vertical use and definitely use the panel technology advertised.

Use a vertical monitor for your code editing. Having twice as many lines of code in front of you at once makes a big difference. Most skeptics haven't really tried it.

Stop wasting time obsessing about the best coding setup and the best editor. I've seen very poor developers with incredibly advanced typing shortcuts that they spend hours perfecting. They still write shit code (maybe they do so faster...). I use the standard configuration for an editor I like, which is a text editor only, not an IDE. I think the extra seconds here and there I spend typing repetitive parts of code give me time to think about what I'm going to do next. If you've ever spent half a day writing macros that save you milliseconds, then you're fooling yourself into thinking that you're increasing your productivity. And as I've mentioned, "lost time" isn't lost if you're spending it thinking about the problem you're solving.

Limit your use of debugging tools. I'm self-taught and in my early years of programming (Pascal) as a teenager I just didn't know how to use a debugger. It's given me a 6th sense in figuring out where the problem is in the code when I encounter a problem. Developing that ability makes you work faster because most times you know immediately where the problem is, instead of hunting things down in the debugger in a systematic fashion. Next time you're tempted to immediately fire up the debugger, just don't and try to figure it out only by looking at the code.

Some countries have special forces operating in Somalia hunting these guys down. Parading and bragging about how much money he's earned makes him a target.

Can't wait for Twitter to join Myspace in the slow death club. This move is certainly a step in that direction.

He should enjoy the money while he's still breathing, which won't last as long as he thinks it will.

The companies providing the APIs don't have to justify the "why". When I was working on Facebook applications a few years ago they would change the API calls and rules almost monthly, app developers had to adapt or die. The changes were often due to apps abusing the system in one way or another, but Facebook defined what was abuse or not. Sometimes the changes just looked completely arbitrary and sometimes meant that some applications were just not possible anymore due to vanishing API calls.

Exporting the private data from Facebook is considered a big no-no, from what I remember the TOS might only allow you to do so for a short period. Maybe that search service gets away with it because it stores an index based on the private data and not the private data itself. If Facebook gets uneasy with how close the index is to the original data or what private information leak might happen if Greplin is compromised, that could be a good enough reason to put a stop to it. Facebook not getting into a private information leakage scandal is probably more important than allowing a startup to live on Facebook data.

Facebook already updates their API and policies very regularly, nothing they give an application access to can be given for granted in the long term. Especially when it deals with private content.

When your service is dependent on another, you're always at the mercy of an update that can shut down the API feature you were using, whether the intention is to kill your service or not.

In this case it seems like the goal of the website is to become a starting point for a person's browsing/searching experience, facebook and google want the same thing. If they feel threatened they could cut just stop Greplin in particular from accessing their API. It's not hard for them to update their API terms of service to make that kind of use forbidden, for instance. That kind of tactic has been used in the past.

That's why I see a lot of uncertainty in the future of that service. I'm not worried about the founder and the company, though, their most likely exit is a talent acquisition by Facebook or Google.

It looks like he's using APIs, which means that he can only index as much data as the services are willing to provide. They can squeeze him out of business by shutting down or reducing access to the data whenever they please. Or just add to their API TOS that keeping a cached copy of the data is a no go.

The service would lose a lot of relevance if suddenly one of the main players like Facebook decided that they don't want to be indexed by Greplin anymore. Without long-term partnerships or contracts with all the stakeholders, I wouldn't bet on that horse.