248
every worthwhile link on how to implement your own squeeblerizer is dead and approximately 40% archived
(files.catbox.moe)
cross-posted from: https://lemmy.blahaj.zone/post/20883609
show transcript
compiler-specific posts:
every software is like. your mission-critical app requires you to use the scrimble protocol to squeeb some snorble files for sprongle expressions. do you use:
- libsnorble-2-dev, a C library that the author only distributes as source code and therefore must be compiled from source using CMake
- Squeeb.js, which sort of has most of the features you want, but requires about a gigabyte of Node dependencies and has only been in development for eight months and has 4.7k open issues on Github
- Squeeh.js, a typosquatting trojan that uses your GPU to mine crypto if you install it by mistake
- Sprongloxide, a Rust crate beloved by its fanatical userbase, which has been in version 0.9.* for about four years, and is actually just a thin wrapper for libsnorble-2-dev
- GNU Scrimble, a GPLv3-licensed command-line tool maintained by the Free Software Foundation, which has over a hundred different flags, and also comes with an integrated Lisp interpreter for scripting, and also a TUI-based Pong implementation as an "easter egg", and also supports CSV, XML, JSON, PDF, XLSX, and even HTML files, but does not actually come with support for squeebing snorble files for ideological reasons. it does have a boomeresque drawing of a grinning meerkat as its logo, though
- Microsoft Scrimble Framework Core, a .NET library that has all the features you need and more, but costs $399 anually and comes with a proprietary licensing agreement that grants Microsoft the right to tattoo advertisements on the inside of your eyelids
- snorblite, a full-featured Perl module which is entirely developed and maintained by a single guy who is completely insane and constantly makes blog posts about how much he hates the ATF and the "woke mind-virus", but everyone uses it because it has all the features you need and is distributed under the MIT license
- Google Squeebular (deprecated since 2017)
Of the available options, this is easily the best, since I can use my own compilation flags to tune the library for my specific target architecture/CPU which can possibly change as the deployment profile for the business case evolves. Assuming it's OSS, I can also fork and adjust the library itself for said "mission-critical" use case.
Also, the Google product being deprecated since '17 is too real 😅...
And here I would argue that the Rust library is strictly better, specifically because it will come with an automated or precompiled build of the C library. Compiling C is such a pain.
Also, on Gentoo Linux, there will be an ebuild that integrates all of the cmake options into the rest of the packaging system and manage the dependencies
Any similar system for Kubuntu 24.04 LTS noobs/normies like me? I don't know what "ebuild" is, but it sounds cool (of course, I could look it up, but I thought I'd just ask).
I'm not a dev-ops dude, but for work, I develop parametric CAD solutions and generative DNNs for CAD. Lots of linear algebra and Pytorch on the GNU-Linux side; lots of Grasshopper for Rhino8 on the Win11 side. Hence, I use Docker to separate my experimental build environments from my production ones.
I've been kinda maintaining my shit "by hand", so to speak, for years now, and I think I'm ready for some automation in that regard.
An ebuild is a definition - a recipe, if you will - of how a package is built from some source by portage, a Gentoo package manager.
Very few things are trully impossible in linux land, but having multiple package managers on a single system is just asking for trouble.
You could try setting up a gentoo prefix and get the benefits of portage that way, but I've not beem able to accomplish that the single time that I tried.
Yup. If source is not available I’m not using it if I have any choice in the matter. Binary distribution is nice, but I’d rather have source.
Plus I’m sure some kind soul has created a build pipeline that autogenerates binaries from the source. I can always either use that or clone and customize it. It’s a natural separation—as a dev I’d like my responsibility to end at “I merged working code to trunk”.