@gisgeek@floss.social
@gisgeek@floss.social avatar

gisgeek

@gisgeek@floss.social

Debian Developer, GIS geek, FLOSS activist. Researcher and data scientist.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

tripleo, to random
@tripleo@fosstodon.org avatar

All you nutcases still using , what's actually wrong with it?

aka What are the sharp edges?

gisgeek,
@gisgeek@floss.social avatar

@leonerd @tripleo I can only confirm this, even if currently FFI support does allow to bind #Perl to C/C++ quite easily. I find the general situation much better of other seasoned languages such as #Tcl or #Lisp, instead.

bagder, to random
@bagder@mastodon.social avatar

"To me, the latest is the latest my OS provides me. If maintainers dont care about pushing the latest into the OSes they support, it's not me to blame. I think curl maintainers should push Centos to provide the latest to all users. What's the purpose of you fixing multiple bugs and security holes if you dont spend time to make it available to the broader audience?"

We are obviously all just too lazy.

https://github.com/curl/curl/issues/13546

gisgeek,
@gisgeek@floss.social avatar

@bagder well, Debian Trixie/testing comes with 8.7.1, currently. But of course, stable has 7.74.0. maybe it depends on what is the distro target.

gisgeek, to python
@gisgeek@floss.social avatar

In recent years, many geodata providers have been exploited by multiple libraries - that use their REST APIs - which are usually written in . That's the case of multiple archives. Most of them use the same multi-threaded '' and '' packages below. I wonder how all of them may suffer the same problem of hanging forever on massive downloads. Come on, guys, it is not rocket science. That's really annoying...

sesivany, (edited ) to random
@sesivany@floss.social avatar

If you want your posts to be visible, use hashtags. This is more important here than anywhere else, because there is no cross-instance full-text search. And people do follow hashtags.

In my experience well-hashtagged posts get half of their interactions from people who don't follow me.

gisgeek,
@gisgeek@floss.social avatar

@sesivany That's true and it remembers me the old good times of topics.

jwz, to random
@jwz@mastodon.social avatar

Today in "Daylight Savings Chaos Monkey"

If you have backups but you have never tried to restore them, you do not have backups. This is true.

I made a MySQL oopsie, and tried to restore a table from my voluminous backups of same, and was met...
https://jwz.org/b/ykM4

gisgeek,
@gisgeek@floss.social avatar

@jwz and now with the general move of time_t to 64 bits I guess a lot of historical backups will become unusable.

deshipu, to random
@deshipu@fosstodon.org avatar

Contributing to open source projects I increasingly feel like a bee watching a bunch of bears fighting over the honey, trying to not get trampled underfoot, and thinking about how to rebuild at least some of the the stuff that we rely on once the corporations lose interest and move on.

gisgeek,
@gisgeek@floss.social avatar

@deshipu Just avoid the hype stuff, i.e. do not follow trendy projects, they are often traps for bees. There are really a lot of projects out there that need help and it is a pleasure to work on.

gisgeek, to random
@gisgeek@floss.social avatar

@debacle @z428 I only use upstream packages for some key desktop apps. About screencast and streaming, wayland Is still quite challenging in stable, needs some edge version and probably patches to work. Indeed I still use Xorg also due to Nvidia support.

gisgeek,
@gisgeek@floss.social avatar

@z428 @debacle The stay-current hype in development is an anti-pattern for enterprises (and even any user that needs to work, not play with toys) that I don't share.

Unfortunately, this is not a Linux-only issue and even the reason why I'm becoming quite choosey in my tooling, starting from languages. A library. package or tool that has a 1-2 years lifetime - as many in the Javascript/Python/Rust ecosystems - are not for me.

Desktop stuff is not different in those regards.

gisgeek,
@gisgeek@floss.social avatar

@z428 @debacle Statically built binaries can be run on Linux forever, but sure: if the program is a dynamically linked one, or it depends on certain interpreted languages and their libraries, the possibility of problems after 10 years is concrete. That's mitigated in the proprietary world by collecting together all dependencies in a tree. I think the same can now be done better by containers or other tools, such as and . POSIX standard does not prevent problems in such cases.

gisgeek, to foss
@gisgeek@floss.social avatar

A due to myself. I'm an old-style with more than 30 years of experience with multiple flavors of *nix, networking, , and development. I'm a researcher and data scientist in the and domains, with a background in and . I'm also a Developer, a full-stack dev, and a system administrator, because of my geekness. I'm a type of CS professional that is currently almost extinct, I would say...

danjac, to rust
@danjac@masto.ai avatar

Managed to get through most of the book over the weekend - I tend to jump back and forth a lot in learning, so some things not completely understood yet (e.g. lifetimes and wtf 'static means).

Some thoughts - the Rust type system and compiler will really do their best to stop you breaking shit, and that is IMHO a good thing.

Most of my career is spent in maintenance - bug fixes & features. Speed of writing code is never the problem. Speed of fixing it is.

gisgeek,
@gisgeek@floss.social avatar

@danjac I don't know Rust, but I guess a good knowledge of assembly and computer architectures help a lot to truly understand the language, on the basis of what I read as the static description in Rust.

gisgeek, to random
@gisgeek@floss.social avatar

My first on *nix was on ~1993 (I think). In 1995 I moved to and never changed. So I will turn 30 years of use next year. Wow!

qlp, (edited ) to debian
@qlp@linh.social avatar

Okay, I took the jump from stable to testing on my Debian DevTest laptop.

What could possibly go wrong? 😄

gisgeek,
@gisgeek@floss.social avatar

@qlp I used sid for years as my day-to-day box without significant issues. In many cases, it is even less broken than testing, where transitions and RC bugs could prevent some packages from working for long periods. Nowadays I simply prefer using stable with VMs and chroots instead.

gisgeek,
@gisgeek@floss.social avatar

@qlp On sid/testing installing apt-listbugs and never using full-upgrades help a lot. But yes, oddities could happen from time to time.

gisgeek, to Ansible
@gisgeek@floss.social avatar

Let's start 2024 with a new series: Tool to consider that could deserve your attention.

Most of you use for devop operations.

Meet , a light framework to perform devop operations, which can be easily customized for multiple use cases. And it's not a single-company show, which is a nice plus.

https://www.rexify.org/

gisgeek, to scheme
@gisgeek@floss.social avatar

Good intentions for the new year. Deeply dive into new (for me) niche languages with a lot of interesting programming patterns. I'm now evaluating a couple of #Scheme implementations: #Racket and #Guile. Hints are welcome.

gisgeek, to linux
@gisgeek@floss.social avatar

I'm old enough to have begun using before , and in the first years, I used *nix (well, SunOS, Solaris, and Digital OSF/1, to say more precisely) for so long. I'm what nowadays is considered a Veteran Unix Admin or . I'm still curious enough to stay updated about current tech, but I wonder how many people out there in the are still passionate about tech novelties but even cultivate legacy knowledge such as #C, , , and , and above all why?

gisgeek, to debian
@gisgeek@floss.social avatar

After many years of package maintenance in , I finally resolved to adopt the orphaned tool for , which is based on and we actively use at work. It used to be also packaged in Fedora until 2015.
See https://github.com/fpl/autodir for the new home and updates.

In the meantime, I also updated its seasoned howto, but I wonder if the Linux Documentation Project will live again or not in the future: apparently the whole project pages and mirrors seem frozen as an ice cube.

sesivany, to ai
@sesivany@floss.social avatar

I wonder what will happen when most of the content that generative learns from is AI-generated bogus content. Will we call it degenerative AI? 🤔

gisgeek,
@gisgeek@floss.social avatar

@sesivany, that's the risk of continuous learning. Drifting is something that one could expect from ML models, so a continuous screening of the data population used is also expected.

gisgeek, to random
@gisgeek@floss.social avatar

multi-threaded script. Do suggestions about how to debug such type of issues? Maybe in the main thread? Of course that does not happen always.

paulox, to opensource
@paulox@fosstodon.org avatar

I didn't know about this SQLite fork. ⎇

I wonder if it can be used with the same SQLite backend for the Django ORM. 🤔

You can read more about our goals and motivation in the announcement article 👇
https://itnext.io/sqlite-qemu-all-over-again-aedad19c9a1c

gisgeek,
@gisgeek@floss.social avatar

@paulox You should consider that SQLite can be extended with contributed shared libraries. For instance, Spatialite is a plain SQLite with OGC spatial add-on extensions. So no need to fork anything. This is the same approach used at the time in Tcl where multiple extensions have been created, such as Expect language and others.

gisgeek,
@gisgeek@floss.social avatar

@paulox This is explained in https://sqlite.org/src/doc/trunk/README.md, and of course, nowadays all that could seem anachronistic. I see exactly the same approach elsewhere.

It makes some sense when a project is based on the work of very few people at the origin), because it solves a problem from the root: governance. Other people are included in the core team on invitation. Other people can only provide proof-of-concept patches.

gisgeek,
@gisgeek@floss.social avatar

@paulox of course, in the long term, the sustainability of projects with this kind of governance is questionable. Nowadays, if I look at governance problems of some big FOSS projects, that model has some attractiveness. 😎

vwbusguy, (edited ) to ruby
@vwbusguy@mastodon.online avatar

Some have said that I am capable of writing in any language. Behold my .

gisgeek,
@gisgeek@floss.social avatar

@vwbusguy 😆 regex rocks

stefano, to linux
@stefano@bsd.cafe avatar

Monday morning, the peak time for requests and issues. One of the things I try to avoid on servers is using external repositories for installing specific package versions. Just a while ago, a developer asked me to install the php module for on an (old) server with an external php repository. Even though the server is being decommissioned, this is a temporary operation for migration. Unfortunately, I had to tell them that the packages are no longer available because that repository no longer provides them. Tech debt always comes knocking sooner or later.

gisgeek,
@gisgeek@floss.social avatar

@stefano something difficult to explain to too many devs and users is that language specific package systems (call them npm, gem, cpan, cargo, pip or whatever) do not solve the issue, only hide the whole chain of dependencies, often in sneaky manners. You will pay what's due sooner or later.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • megavids
  • rosin
  • khanakhh
  • osvaldo12
  • Durango
  • mdbf
  • DreamBathrooms
  • ngwrru68w68
  • magazineikmin
  • thenastyranch
  • cisconetworking
  • Youngstown
  • slotface
  • kavyap
  • JUstTest
  • cubers
  • anitta
  • GTA5RPClips
  • tester
  • ethstaker
  • modclub
  • InstantRegret
  • everett
  • provamag3
  • Leos
  • normalnudes
  • tacticalgear
  • lostlight
  • All magazines
    •