The Changelog: Software Development, Open Source - No Maintenance Intended (News)

Episode Date: March 18, 2024

A new badge for open source projects that won't be getting any maintenance, everything Chip Huyen learned from looking at 900 open source AI tools, CNBC writes up tech's renewed layoff trend, Teable is a Postgres-Airtable fusion & Target announces an open source fund.

What up nerds, I'm Jared, and this is Changelog News for the week of Monday, March 18th, 2024. Did you know that poet Shel Silverstein predicted chat GPT all the way back in 1981? I present to you the homework machine. The homework machine. Oh, the homework machine. Most perfect contraption that's ever been seen. Just put in your homework, then drop in a dime, snap on the switch, and in 10 seconds time, your homework comes out quick and clean as can be. Here it is. Nine plus 4, and the answer is 3. 3? Oh me. I guess it's not as perfect as I thought it would be. Okay, let's get into the news. No maintenance intended. When you really think about it, open source is a gift to the world. Some gifts are ongoing efforts,

sure, but other gifts are just one-offs. You build something, you give it away, and that's it. You have no intention of improving it, maintaining it, or even looking at it again. That's totally fine, but if and when that's the case, it's a good idea to clearly communicate that expectation. Enter the no-maintenance-intended badge. Add the badge to your hobby project and send folks to this message when they follow the link. Quote, if you're here, that likely means a project linked you here. Thanks so much for being interested in that project. Open source is rewarding, but it can also be exhausting. The linking projects code is provided as is and is not

actively maintained. The authors of that project invite you to peruse their code and even use it in your next project, provided you follow the included license. No guarantee of support for the code is provided, and there is no promise that pull requests will be reviewed or merged. It's open source, so forking is allowed. Just be sure to give credit where it's due. End quote.

In a somewhat ironic twist, the no-maintenance-intended- intended badge project itself is maintained and is on GitHub. What I learned from looking at 900 of the most popular open-source AI tools. Four years ago, Chip Huynh did an analysis of the open-source machine learning ecosystem. Since then, the landscape has changed, so she revisited the topic. This time, she focused exclusively on the stack around foundation models. This is a hashtag long read, so I'll just list her personal favorite ideas for you here and let you take the deep dive on your own time. 1. Batch inference optimization like FlexGen and Llama.cpp

2. Faster decoding with techniques such as Medusa and lookahead decoding. Three, model merging with MergeKit. Four, constrained sampling with outlines, guidance, and SGLang. And five, seemingly niche tools that solve one problem very well, such as INOPs and safe tensors. You can find links to all these in our companion newsletter, and you'll also find most of them on her cool LLM repos list, which is also in our companion newsletter. It's now time for sponsored news. If you're experiencing slowdowns in your application due to excessive joins and lengthy query durations, it might be time to reevaluate your database. Have you considered

the power of a graph database? GraphDBs let you model data the way it looks in the real world, instead of forcing it into the confines of traditional row and column structures. Move beyond the limitations of relational databases for tasks they aren't built to handle, whether it's managing intricate supply chains, detecting fraud, conducting real-time analytics, or powering Gen AI applications, graph technology excels in scenarios rich with interconnected data. Neo4j offers the flexibility to develop using your preferred programming languages and connect via any driver, ensuring

seamless integration with your existing stack. People are solving some of the world's biggest problems with graphs. Now it's your turn. Begin your journey at neo4j.com slash developer. Once again, that's N-E-O number four, j.com slash developer. Check it out. And thanks once again to Neo4j for sponsoring Changelog News. Laid-off techies struggle to find jobs with cuts at highest since 2001. Alex Kohler, writing for CNBC, quote, Since the start of the year, more than 50,000 workers have been laid off from over 200 tech

companies, according to tracking website layoffs.fyi. It's a continuation of the predominant theme of 2023, when more than 260,000 workers across nearly 1,200 tech companies lost their jobs. Alphabet, Amazon, Meta, and Microsoft have all taken part in the downsizing this year, along with eBay, Unity Software, SAP, and Cisco. End quote. Rolling into the new year, it seemed like sentiments were shifting positive again, but what momentum we had gained seems to have dried up.

Adding to the layoff-based trepidation is the pending CodeGen AI takeover, of which Devin is the newest poster child. As DHH wrote on the subject, developers are on edge, but I'll echo his concluding words, because they're as good as anything I've got on the matter. DHH says, So while it's hard to do, it's useless to worry. The future is out of your hands and out of your control. No profession has ever successfully resisted automation or redundancy in the face of technological advancement over the long term. Screaming at Devin will only distract you from enjoying the last glorious years of a golden run. T-Bull is a Postgres Airtable Fusion. T-Bull is a super fast, real-time, professional, developer-friendly,

no-code database built on Postgres. It uses a simple spreadsheet-like interface to create complex, enterprise-level database applications. Unlock efficient app development with no code, free from the hurdles of data security and scalability. End quote. Do you like the idea of Airtable, but would rather it be self-hosted and actually just Postgres under the hood? I sure do. Announcing Target's open source fund. Brian Muenzenmeier on Target's tech blog says, quote, it's no secret that an engaged open source team yields transformative, innovative, and collaborative outcomes. As we look to discover more ways to engage our team in Open Source,

the success of Sentry's and Microsoft's Open Source fund efforts inspired us to pursue a similar program. Our efforts today, along with sustained and empowered upstream engagement, aim to strengthen the Open Source movement, reduce maintainer burnout, and normalize corporate contributions back into the ecosystem. We learned a lot from our first round and look forward to iterating with our community on future rounds, end quote. The good news is that Target has seen the light and is now donating to five open source projects selected through their process. Kudos to them for that. The bad news is their 2024 donation budget is $12,500,

which will hopefully move the needle for the projects selected, but doesn't seem like much from a publicly traded company with a market cap of $75 billion. That's the news for now, but also scan the companion changelog newsletter, link in the show notes, which includes more awesome stories like the Posse Pulse, Marker, an open source desktop app for markdown editing, and more. We have some awesome episodes coming your way later this week. Chris Moore talking True Naz and the end of the free BSD version, and Cameron Say from last year's Mainframes are still a big thing. Have a great

week, leave us a five-star review if you dig our work, and I'll talk to you again real soon.