HNHacker News
TopNewBestAskShowJobs

ISV_Damocles

571 karma · joined September 26, 2012

submissionscomments
ISV_Damocles··on Show HN: Marsha – An LLM-Based Programming Language
So, targeting other languages is on our roadmap, and we have found higher reliability when we stick to a functional style with Marsha, so I think Haskell will be one of the "good ones" amongst target languages, eventually.

But we are focusing on Python first because it's one of the most popular languages on the planet and many LLMs out there are able to generate high quality code for it, while less-used languages tend to produce worse results. Haskell may or may not need a manually-tuned LLM to generate solid results.

ISV_Damocles··on Show HN: Marsha – An LLM-Based Programming Language
Simply, Marsha is not Python. ;)

More seriously, in Python, types are optional and the exact behavior is explicitly defined at runtime, but the author of the code already had a behavior in mind when they wrote the code. Getting the intended type information from the user to provide to the LLM improves the quality of the generated code, which is actually somewhat lossy. The "intended type information" isn't "explicit type information", though. We support any character besides these four as part of a type `(),:`. That includes spaces and periods, so you can describe the type as a short sentence if you aren't exactly sure what the structure ought to be.[1]

And since in Marsha you don't actually write an imperative function body, argument names don't matter, only the argument order (as Marsha does not support named parameters, at least right now) so there's no argument naming in the declaration, which can make it more succinct than Python def lines in certain situations, though to be fair none of our examples do this.

[1]: https://github.com/alantech/marsha/blob/main/examples/web/du...

ISV_Damocles··on Show HN: Marsha – An LLM-Based Programming Language
Well, to be fair, we do bulk up what you type quite a bit[1] to improve GPT's response.

[1]: https://github.com/alantech/marsha/blob/main/marsha/parse.py...

ISV_Damocles··on Show HN: Marsha – An LLM-Based Programming Language
yeah, the AI hype cycle can be a bit tiring, but I'm glad you took a second look. What parts did you find interesting, in particular?
ISV_Damocles··on Show HN: Marsha – An LLM-Based Programming Language
Btw, here's the test job: https://github.com/alantech/marsha/blob/main/.github/workflo... And the core script for the job: https://github.com/alantech/marsha/blob/main/marsha/.time.py
ISV_Damocles··on Llama 2
So this comment inspired me to write a Roman Numeral to Integer function in out LLM-based programming language, Marsha: https://github.com/alantech/marsha/blob/main/examples/genera...
ISV_Damocles··on WebAssembly Tail Calls
It would help any recursively-defined code, parsers of some input format are often more naturally represented that way, for instance, where rewriting as an imperative loop requires mixing of different logical states, eg, a JS JSON parser might determine if the current character is for an object, array, or leaf type, and then pass the input along to whichever that is, and if it was an array, for instance, would then pass back to that initial function one character over to determine what the first type of the first element of the array is. This is easy in a recursive style and would require all of this logic to be mixed in the same function with some state tracking variable to decide which branch to use in imperative style.

You can use trampolines[1] to get around this, but there's a performance penalty versus TCO and other recursive optimizations.

[1]: https://blog.logrocket.com/using-trampolines-to-manage-large...

ISV_Damocles··on Show HN: IaSQL beta – cloud infra as data in PostgreSQL
Another co-founder of IaSQL here. There are definitely some similarities with IaC tools in IaSQL; at it's core there's a diffing system between the cloud infrastructure state and the desired state that is used to execute AWS SDK calls to update your infrastructure.

But we also use the same mechanism to take changes in your infrastructure, perhaps made via the AWS console or an IaC tool like Terraform, and pull that into your IaSQL database. We use an audit log tracking who made changes where to determine which way the synchronization process goes, and this gives us a lot of "powers" over normal IaC.

First, you can inspect the tables in IaSQL and be sure that it's actually the state of our cloud (within a few minutes of difference), so you can be more sure that the changes you're writing up are likely to work.

Second, the changes you write are SQL statements to mutate your infrastructure rather than editing a declaration file, which is a different mental model vs IaC. But you gain from the constraints and foreign key relations on tables to catch mistakes in configuration before execution even happens to a greater degree than IaC tools can as this is like a "type safety" on top of dynamic information (other tables you depend on).

Third, you can enter an IaSQL transaction (slightly different from a normal Postgres transaction as you can't hook function calls on `BEGIN` and `COMMIT`) to make these changes and inspect what IaSQL will do like you can with IaC tools, but if there is a failure, the `iasql_commit` will take the information from the audit log and roll back the changes that were applied by applying the prior versions in reverse, so IaSQL should always be in a working state.

Finally, (since this is getting long) the live-infra data model of IaSQL makes dynamic infrastructure much simpler (eg, you deploy isolated instances of your services per customer) -- anything with a SQL client and credentials for the IaSQL database can connect and initiate these changes, which can be Postgres stored procedures you write and added to the database.

We're in beta so our coverage is not as large as existing IaC tools, but we now cover 25 different AWS services with our modules: https://iasql.com/docs/modules/ But please let us know if there's something you use that's missing. :)

ISV_Damocles··on ELIZA is Turing Complete
Well, no, because it's not December yet. Pretty sure it's just a typo on the ISO date format.
ISV_Damocles··on Exponential Smoothing: faster and more accurate than NeuralProphet
https://dictionary.cambridge.org/grammar/british-grammar/nei...
ISV_Damocles··on I've been abusing HTTP Status Codes in my APIs for years
The immediate rebuttal to this whole article is that you can have custom response bodies on any status code, not just when the status code is 200, so you can satisfy both kinds of users by having `/api/v1/employees/100` with `404` as the status code and `{ "success": false, "reason": "No employee with ID 100 exists" }` while `/api/v11/employees/1` can fail with `404` as the status code and `{ "success": false, "reason": "No such path: /api/v11", }`

Now for those who are exploring your API with curl or are reading logs from their service they can see the human-legible information that may help them debug the error, while their application logic can continue to use `2xx` for success, `4xx` for my own failure and `5xx` for a server failure that I should retry after some back-off time, allowing it to do the most sensible thing it can in the situation, rather than explode in an uncontrolled fashion when an endpoint the developers only got successful 200 responses from during development suddenly encounters a 200 response with an error payload.

ISV_Damocles··on Linus Torvalds: Rust for the Kernel Could Possibly Be Merged for Linux 5.20
Both of statements can be true, though. Developers skilled enough / brave enough to write kernel drivers can be very careful about not leaking memory or mismanaging the hardware in question, and Rust can make that job much easier by being pedantic about memory mistakes or type misuses (with custom types that are validated at compile time but no runtime overhead).

I've worked on C and Rust codebases, though nothing near as complex as the Linux kernel, and I'm excited for this change as it will reduce the burden on the kernel developers to reach the same quality output, so we either get drivers faster, or drivers for hardware we wouldn't have otherwise received, or both.

As far as the fear of Rust kernel drivers that don't compile with GCC's Rust frontend, I trust Linus to keep the bar high on letting this feature in, and that kernel drivers accepted into the tree compile with just GCC as they always have.

Third-party drivers may not, but that's also true today if someone wrote a driver that only compiled with LLVM. This doesn't really happen in practice as it goes against the path-of-least-resistance for the developer: they'd have to switch their toolset out when working on that driver vs everything else, and the auto-building by DKMS would probably fail on them if they used it during development -- all of the conventions used by the rest of the kernel infrastructure will keep them in line.

ISV_Damocles··on Ask HN: What do you use VMs for regularly?
I am pointing at GNOME because from my experience you can basically substitute any of the big distros these days and get a near identical experience.

What webcam, mouse, or keyboard doesn't work with Linux? I have not run into one, and I have several weird ones like the Tap keyboard.

ISV_Damocles··on Ask HN: What do you use VMs for regularly?
No, I hate debugging on personal machines. I really haven't had GPU driver issues for 10 years on any of my Linux machines, and it's a pretty eclectic mix of Ubuntu, Fedora, Manjaro, and now one Steam Deck on SteamOS.
ISV_Damocles··on Ask HN: What do you use VMs for regularly?
No judgment. Same reasoning for me. In 2016 the bugs with MacOS got annoying enough that I looked around again and I had a more stable experience with GNOME 3 on a Dell laptop than anything on MacOS or Windows and switched.

I had way more problems debugging weird crap with the Wifi when walking between meetings with my MacBook Air, annoyed that when I plugged back into my monitor at my desk I would have to rearrange my windows on my desktops again (GNOME 4 puts the windows back where they were if you plug back into the exact same monitor). And dealing with Homebrew BS because Apple likes to break it on OS updates and sometimes it would break randomly on its own while Debian-derived distros never have package issues pushed me back.

Just saying that you really ought to try it out again, and Fedora is a super-polished GNOME experience if you want "Mac-like, but not annoying."

ISV_Damocles··on Ask HN: What do you use VMs for regularly?
Sorry you've been unlucky, but I've bought brand new Dell and Lenovo laptops and slapped Fedora on them and they just worked with no driver BS involved. Literally did this just a few months ago.

Ubuntu/Debian tend to take longer to support brand new hardware because they're more conservative with kernel updates, but you can also just use the System76 PPA to get a super-recent kernel on Ubuntu. But faster-paced or rolling distros like Fedora or Manjaro are best for that kind of thing.

I haven't had a single GPU-related issue, AMD or nVidia, for about a decade, if you're willing to go with non-free drivers.

ISV_Damocles··on Ask HN: What do you use VMs for regularly?
Will have to "agree to disagree" on your conclusion here.

GNOME 4 is superior to Windows and MacOS as low-friction, high-level OS in my experience. Control panel is clear, if spartan, three finger gestures stolen from MacOS "just work", and the desktop just stays out of the way so I can focus on the applications that I really care about.

I have not had a "struggle with drivers" for over a decade with Ubuntu, Fedora, and Manjaro (YMMV in other distros of course, but it's much more uniform these days), and Windows has become even worse with doing unnecessary BS work in the background slowing down the GUI such that heavyweight desktops like GNOME and KDE feel as svelte as LXDE in comparison.

I honestly can't stand any time I have to use a Windows machine due to how inconsistent the settings are, how many ~200-300ms pauses happen randomly, especially when I'm trying to type, and applications that don't "just work" but only because of the high marketshare I can google a solution. (MacOS is less viscerally disgusting to me, but more disappointing due to all of the "small" bugs that have creeped in despite having 100% control of the hardware in question. Wake-from-sleep shouldn't screw up the Wifi, unplugging and plugging back in an external monitor shouldn't shuffle the window locations, etc.)

ISV_Damocles··on Ask HN: What do you use VMs for regularly?
WSL2 is a VM, though. It was only WSL1 that tried to do a kernel-level shim for the Linux userspace layer.
ISV_Damocles··on Supabase (YC S20) raises $80M Series B
I've personally found Postgraphile to be fantastic. Nicer to use than Hasura and fully OSS: https://github.com/graphile/postgraphile/
ISV_Damocles··on Nexe: Create a single executable out of your Node.js apps
I've personally used `nexe` before and ended up swapping it out for Vercel's `pkg`[1]. `pkg` worked better with building ARM binaries and with less gotchas with multiplatform builds, in my experience.

[1]: https://github.com/vercel/pkg

ISV_Damocles··on UPDATE iasql SET source = 'open';
We're excited to open this up to the world! Let us know what you think. :)
ISV_Damocles··on Burn My Windows
If you use KDE wobbly windows are built in to its KWin compositor. :) I forget the exact details of how to get to it since I don't use KDE anymore (long-ish story), but even most of the silly things Compiz did are a configuration checkbox away in KDE.
ISV_Damocles··on Show HN: QuickServ – User-friendly web server
Will edit this comment with more notes as I review it.

First thing I noticed is that Raspberry Pi / Other OS have the example dump into `temp.py` and then indicate that you should request `test.py` (while Win/Mac examples correctly have them create a `test.py` file), so that could be an issue for novices. (Novices running Linux? My son, for one ;) )

EDIT 1: Compile from Source has this issue, too.

EDIT 2: The bit about `directory/index.ext` definitely matches a lot of existing tooling, but not sure if it belongs in a tool that wants to avoid 20+ years of Web cruft? Extension-less files ought to work for the same outcome, right? Or alternatively if the extension would be unambiguous (eg `calculate.*` only matches `calculate.py` it could just be omitted?

EDIT 3: Handwriting query parsing code is not great. If you're not exposing explicit HTTP verbs (GET, POST, etc) or any of the HTTP headers (since I don't see any logic for handling the headers in the stdin) it's inconsistent to have the users hand-parse the query string or body, especially when it's error prone enough that you have more comments than code in that example. It may make sense to just point them at the Python's urllib library (particularly `urllib.parse.parse_qs` for this example to get a basic dictionary to work with).

EDIT 4: I just skimmed through three of the examples (a bash one, a C one, and a TS/Deno one). Besides the custom parsing logic in all of them, and keeping the "rapid prototyping / hacking" mentality of the project in mind (which I do agree is a good thing), what keeps nagging me is: "what's the off-boarding strategy?" Since it is not intended to be secure for real deployments, is it just "rewrite it in your language of choice"? It would be nice if there were examples of what to do if your prototype is ready to be converted into a "real" project with testing, security, etc in mind, even if it was just an essay on all of the stuff that QuickServ didn't do that a real web server needs to handle. Perhaps in the future user-contributed guides for particular languages and frameworks (eg "Convert your Python QuickServ prototype to Django") could also be added to help them move on more easily, since paradoxically for this project the easier it is to migrate away, the more likely people will use it.

ISV_Damocles··on Mozilla has defeated Microsoft’s default browser protections in Windows
Just a side note: I already can play Halo on Steam on Linux with only one minor issue (Proton's conversion of DirectX to Vulkan works almost flawlessly, but doesn't seem to handle single-faced triangles, so the hologram of Cortana in the opening scene looks bad as you see the back of her head transparently through her face).
ISV_Damocles··on Containers Don't Solve Everything
The heat death of the universe?
ISV_Damocles··on Ask HN: How did you become a software engineer?
M 37

> How did you first get into programming? Were any books, blogs, forums, or people particularly helpful for your growth? What is it about coding that got you hooked?

My family got our first computer when I was in middle school. A Pentium 90MHz with 14MB of RAM, a 200MB hard drive, SoundBlaster 16 sound card, 5.25" and 3.5" floppy drives, 4x CD-ROM drive, running MS-DOS 6.22 and Window for Workgroups 3.11. There was no consumer-facing internet at that time. Just figuring out how to play a game with a joystick required learning about CPU interrupts and DMA channels for the sound card (and telling games which to use for what), serial ports for the joystick, and adjusting memory allocation in `config.sys` and `autoexec.bat` for DOS because none of those things were done automatically back then and different games had different needs, particularly with the memory allocation, so making special 3.5" floppy boot disks was also common.

All of this required learning how to navigate on the DOS command line and editing text files by keyboard only, so it was a pretty high activation energy to play games on the computer instead of my Game Boy, but the difference in the experience drove me to get there.

One day I randomly stumbled upon `qbasic` and the `gorillas.bas` file, a fully working game of two giant gorillas standing on top of skyscrapers throwing exploding bananas at each other, that I could edit and run. It had a few global variables at the top with explanations of what they were for. I recall laughing at changing the gravity to 100x Earth's and watching the CPU player try to throw the bananas and having them drop instantly and explode itself.

From there I tried to make changes to the rest of the code below the commented constants but that usually ended in disaster. I eventually figured out the parts of the code that were used to draw rectangles, lines, and circles, as well as the random number generator, and I ripped them out into separate files and started making "screensavers" that would spam pseudo-random garbage onto the screen in different patterns as fast as the CPU could do it (because I just put the logic inside of an infinite loop, though I think I also had a space bar press check at some point so I wouldn't have to restart the computer to exit).

For as much shit BASIC gets (rightly, for the most part), that immediate feedback and ease of development let me experiment quickly and experience the almost magical ability of computers to turn your imagination into reality if you can make "the right incantation."

I knew that BASIC was too limited to do what the games I was playing could do, and I would need to learn a language like C to get to that point. I was also drawn to seemingly simple things like Microsoft Paint (and later Corel DRAW!) and the simple Sound Recorder application in Windows, and I could not comprehend how these tools were made.

Flipping through some books on Windows programming and OLE and etc in a local bookstore discouraged me because I didn't have the CS background and technical jargon knowledge to understand most of what I was reading. I projected from the simplicity of DOS to what I was seeing with Windows and assumed in a decade it would be even further along and completely beyond my comprehension, so I didn't push much harder than that in middle school.

In the summer between middle school and high school, my uncle gave us a copy of Visual Studio 6.0, which included VisualBASIC 6.0, which let me make desktop applications that looked like the ones I thought were impossible for me, which got me back into software.

My (public) high school was very forward thinking at the time, and offered electives in C/C++ (it was C but with the silly `cout << "Hello, World!" << endl;` stuff from C++) and web development. The C course helped me finally figure out what pointers were, which helped demystify a lot of stuff. There was also a brilliant and humble kid in my class who was one or two years older than me that was working for a local ISP after school (in 1999) that basically knew all of this inside and out. It left a strong impression on me -- that I had taken far too long to figure these things out and that while I found it very interesting, I probably shouldn't try and "compete" directly as a developer, and should instead choose a field where programming knowledge would give me a boost.

The web class was a combination of learning HTML and Frontpage, but the teacher was very nice and didn't discourage me when I started experimenting with this JavaScript I found while poking around, and I made some sites that would make the text appear a letter at a time in a typewriter font as if it was being typed by someone, I made a Single-Page App in 2000 by putting all of the contents of the site into JS strings and having menu navigation presses replace the `innerHTML` property, etc.

That also made me aware of software development outside of what Microsoft provided to consumers, and I was getting fed up with the Visual C++ books I found in the book store that didn't explain anything but walked you through step by step on how to do things in "Wizards."

Near the end of high school or the beginning of college I found Red Hat Linux and I enjoyed playing with that alternate operating system that I could (theoretically) change any part of it that I wanted to. And I ran across a plain text C programming tutorial centered on GCC and Make that explained exactly what was going on and how a binary was made and how it worked with the operating system, and that's when I was hooked on Linux.

Even after that, I was still pursuing a different field from actual CS. I started with Physics and then switched to Electrical Engineering, even getting a Master's there. Along the way I also learned Perl and CGI scripts for web hosts and used that plus my Linux knowledge to help manage my Research Lab's 2U servers, and write helper tools for parsing all of the measurement data we were making.

I had actually entered grad school in 2007 because I felt like the economy was shaky and I had lucked out, avoiding the 2008 crash where I probably would have been fired during those layoffs. I stayed in grad school until 2012 pursuing a PhD, but a combination of getting married and having a kid, my wife also being in EE (and so our economic eggs being in one basket), and a part-time job at the university working on a CRUD web app for gathering, storing, and approving/denying animal research proposals pushed me to go for programming full time.

Since I had mostly happy experiences with Javascript (pre-jquery), I decided to learn Node.js in my spare time while looking for a job, and I started writing some OSS libraries for Node. Most companies I applied to rejected me immediately from "lack of experience" since I didn't have a CS degree, others would interview me but I was rejected by all except 2. One was a super tiny startup run by a husband and wife where the husband gave me a creepy vibe that I rejected, and the other was an early-ish startup that had traction and high growth, but was also having technical issues and had built a core piece of their technology in Node.js for some reason: Uber.

That was the wildest job I had ever held. The highest highs and lowest lows of my career were all there. I gained a lot of operational experience and architectural knowledge there, helped scale the Node.js-based dispatching system, part of the team that did the second version of Surge pricing, part of the team that built UberPOOL in 3 weeks, built a "hot rod" stream processing system from scratch in Java (the various Apache projects weren't reliable enough, not even Kafka) for the third version of surge, and helped build UberEATS' first and second versions of surge.

The technical side was great, but I was emotionally drained at that point and didn't like what I was seeing from the top, so I left to join another startup that had many technical similarities to Uber but in a different market, but it sold itself to a competitor two years later, so I left.

That's when I co-founded a startup with a friend and colleague from Uber since then, which gets me to the present day. Not going to advertise the startup, we're still too early-stage to have customers, but I've learned a lot about compilers, virtual machines, and type theory in this startup, things that high school me had assumed were beyond my abilities, and I really enjoy what I'm doing and the people I'm working with.

The uncertainty of whether we will be able to reach product-market fit and have the startup actually take off is what I dislike the most about the startup, but it's up to me and the rest of my team to figure this out. That uncertainty is the price that has to be paid for the freedom of choosing exactly what to work on, and I think I'm finally confident-enough in myself to be ready to pay that price.

ISV_Damocles··on Linux distro creator tries to enforce a trademark he doesn't own
Yes, if they were perfectly fine with using that address in public records they should understand that it will be searchable, tying the business to that address.

There are tons of cheap services[1] to provide obfuscated business addresses for LLCs very cheaply, and by the time you're navigating filing documentation for legal trademarks I would expect this to be a known solution, as their legal counsel should've brought it up when they were establishing the LLC in the first place.

[1]: https://www.noobpreneur.com/2018/05/04/how-to-use-a-virtual-...

ISV_Damocles··on Linux distro creator tries to enforce a trademark he doesn't own
That's actually the edited version, the original one stated:

> ...which affords certain protections under US trademark law and has enabled elementary to prevent people from distributing other items under our brand name—but I absolutely do not want to go the legal threat route

Which is much more of an "it'd be a shame if something happened" kind of phrasing.

ISV_Damocles··on Linux distro creator tries to enforce a trademark he doesn't own
The only address I see is the one for an LLC, which is also a public address?

If you mean the email address, it's just your first name and your company, which is the same name you're using for HN and Github? How are they not already publicly associated?

Please, just "mea culpa" that you got off on the wrong foot here, unban the guy, and work out this situation slowly over time. The repo has only 2 stars!

Your ridiculous overreaction is what makes me never want to touch Elementary. I don't want to make this about me, but I'm in your customer profile, typing this on a System76 laptop and having a PineBook Pro sitting next to me. I want to pay OSS companies for quality work! You should really think about what you're doing here and what kind of reputation you want your company to have amongst your potential customers.

ISV_Damocles··on Linux distro creator tries to enforce a trademark he doesn't own
If that was true, you never would have brought up trademarks in the very first post of the issue?

You don't "want" to use the legal threat of trademark law, but you'll dangle it in front to try and coerce your desired behavior.

And there's no doxxing here. That's a legal document he pulled up that's publicly available.

← PreviousPage 2 of 5Next →