Thank you @horgix.fr for sponsoring Conference Hall and Open Source projects. It helps us a lot!
github.com/conference-h...
@horgix.fr.bsky.social
Staff SRE at @payfiteng.bsky.social & Maรฎtre Raclettier โ Mostly posting about conferences & tech โ ๐ง Automation, โค๏ธ Open-Source, โ๏ธ Rust, ๐ฃ๐๏ธ Conferences {organizer, speaker, frequent attendee} โ Mandatory ยซ I use Arch btw ยป
Thank you @horgix.fr for sponsoring Conference Hall and Open Source projects. It helps us a lot!
github.com/conference-h...
Been using it as a speaker a lot, and wish I had it at hand when I still organized conference a few years ago so... it's my pleasure to do so ๐ We're pretty lucky in France to get so many great tech conference, and with such a handy tools combo that Conference Hall + OpenFeedback is!
20.07.2025 22:11 โ ๐ 1 ๐ 1 ๐ฌ 0 ๐ 0Funny how we force engineers to write ADRs for every tradeโoff, yet billionโdollar strategic bets live in hallway whispers and fluffy slide decks.
What if we treated business decisions like architecture? Context, options, tradeโoffs, metrics, all in a repo, versionโcontrolled and reviewed?
๐ฌ Les vidรฉos du #Breizhcamp2025 sont lร !
Tous les trรฉsors du code enfin dรฉvoilรฉs ๐
Pas besoin de fouet, juste d'un clic
www.breizhcamp.org/conference/p...
Ou directement :
youtube.com/playlist?lis...
An interesting undocumented API just hit my automation. Looks like AWS is planning a Vector DB for S3? Or some kind of storage Vector storage offering? Operations include:
- QueryVectors
- PutVectors
- PutVectorBucketPolicy
- ListVectors
- ListVectorBuckets
Oh I can definitely relate ๐
The fact I have an extension to even count tabs is a sign in itself that it went too far a long time ago already. I even had a Prometheus exporter for that at some point in order to be able to graph my tab count and motivate me to clean them up (guess what, it failed)
C'est cool ouaip, mais le truc cool c'est surtout la spec derriรจre je trouve (specifications.freedesktop.org/mpris-spec/l...) !
Je m'en servais avant de passer sur github.com/tramhao/term... (qui d'ailleurs est compatible MPRIS lui-mรชme) qui fait dรฉjร tout ce que je veux dans mon terminal au final
One more day of "damn I love Arch Linux". I get an error while updating approximately once every 2 years - and it happened today. First reflex: check archlinux.org - and here we go without surprise, it's documented on the frontpage with the fix ๐ (archlinux.org/news/linux-f...). 20sec resolution.
02.07.2025 12:42 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0Interested to learn about what it (literally) costs to build a unicorn scale-up such as @payfiteng.bsky.social on top of (a lot of) third parties?
I'll be sharing our bills and analyzing them at @breizhcamp.org this afternoon at 16:00 in Amphi A, and will to my best to wrap the event nicely! :)
Name a more iconic duo than @julientopcu.com & @tpierrain.bsky.social to talk about architecture patterns & modularity at @breizhcamp.org ๐คฉ Let's see what Hive looks like :)
Time to sit back & listen rather than taking notes - if you can't wait for the recording โ bsky.app/profile/juli... I guess!
Last day of @breizhcamp.org opening with a talk on the wizardry of speedrunning Paper Mario!
It's always amazing how far they go to find glitches and exploit them; 5 years of work to find a way to do arbitrary code execution, and so many steps to then do it!
Glad we have @speedons.fr in France :D
Missed my talk today at @breizhcamp.org about Diรกtaxis applied to platform documentation (theory + experience feedback from @payfiteng.bsky.social) ?
While waiting for the recording, here's the content (in ๐ซ๐ท) I presented: s.42l.fr/breizhcamp20...
25min was short, so shoot questions here on Bsky ๐
Welp, that's a wrap, with an on-point closing word, and as usual with @rluta.bsky.social , that was an astonishingly great talk :) Recommend watching to anyone!
Plus, exploring legacy code with such bravery is definitely in itself in the theme (indiana Jones) of @breizhcamp.org this year :D
We're in 2025, you got observability means & tools everywhere:
- Debuguers
- Profilers
- APM
- OpenTelemetry
- Glowroot (glowroot.org - didn't know about it!)
- ...
They usually lack on legacy software, but are a good place to start regaining control
Glad to see the "ops" patterns that can help manage the risk mentioned :D
- Canary deployments
- Feature flipping
- A/B testing
- Traffic mirroring
... help controlling deployments and reducing impact in case the legacy refactoring breaks something.
And vouching for multi-competence teams <3
Don't fear refactoring through exploration and risking breaking things if you _know_ you risk breaking them and are fully ready to rollback.
Actively inline-refactor things to reduce useless parameters while keeping things safe. Your IDE will do everything by itself.
Glad to see "Working Effectively with legacy code" mentioned - it's actually one of the books we have on the shelves at @payfiteng.bsky.social :D
But it's now obsolete and doesn't work in many cases.
"Refactoring" by @martinfowler.com is great - and I actually have it on my own bookshelf!
Generate the missing documentation.
Plantuml and MermaidJS are cool because they generate visuals from text, and it's easy to generate text based on existing code.
The dependency map of IntelliJ helps identifying which parts of the codebase depends on which the most.
It even eases finding the parts that are not relied upon by anything, nor rely on anything, and therefore... can just be removed.
Analysis of version control history leads interesting bits:
- Software life over time
- What part of the system/code were changed the most
- What files the "core historical contributors" touched the most
It allows isolating the bits of the codebase that are most likely the most sound
We have software that handles wider range of cases with better performances than 15 years ago. We also have way better hardware. Some things that were impossible to handle "as-is" before and required layers to mitigate can now be handled "live".
26.06.2025 12:26 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0Only strong recommendation for this talk: know your migration path. Know that you can rollback if you have an issue. 80% of the work will be figuring out that migration path, with usual patterns (e.g. strangler pattern).
Having a "v2" that implies changes on client is usually a hell, avoid it.
Even if some things can seem complex and have been dreading for long (upgrading runtime/platform), they're good candidates to start with - same for config cleanup, build updates, etc.
And these do _not_ require to understand business at first! Purely technical, yet gains
Also, docs.openrewrite.org
Reminder of the 3 axis to evaluate software:
- Functional (features, user doc)
- Operational (perf, security, resiliency, support doc)
- Developmental (tech debt, DORA metrics, dev doc)
I'm a simple man - I see โชa talk with @rluta.bsky.social speaking on the agenda of @breizhcamp.org, I go there :D
Let's find out some tips on how to evolve legacy software painlessly!
I frequently share his talk on performance (www.youtube.com/watch?v=nEye...) to people - hopefully this one too :)
Interested to learn about Diรกtaxis (diataxis.fr) and how we applied it to our internal developer platform's doc (and more) at @payfiteng.bsky.social ?
I'll be presenting it this afternoon at @breizhcamp.org at 17:35 in Amphi E, and I'll keep you awake for the last talk of the day !
Arf, si j'avais vu plutรดt j'aurais pu t'en ramener un ร BreizhCamp, j'en ai en rab' :)
Si jamais tu ne le retrouves jamais et que tu en veux un, n'hรฉsites pas ร faire signe, je dois avoir รงa quelque part !
Oh, and I definitely have to check github.com/Nuclear-Squi... out ๐
26.06.2025 09:26 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0