Romain Manni-Bucau

Congrats on the site revamping! (and the tool but I already knew this one ;)).

[command] add auto-logs command by rmannibucau · Pull Request #2 · yupiik/yuc

My modest attempt to make it easier to work with multistack and inconsistent #json logging formatters (#java, #dotnet, #go there)

-> github.com/yupiik/yuc/p...

kubectl logs whatever | yuc auto-logs

and enjoy :), no more headache to recall the jq pattern/syntax to use or switch between 5-6 aliases

Finally there is some specific UI for #openrpc a bit more advanced/usable than the default playground, no more need of the #swaggerui dirty hack (using path as method)

> octanolabs.github.io/d0x/#/etc

Still hoping @intellijidea.com gets support for #java #annotationprocessor with its incremental compilation optimization

> youtrack.jetbrains.com/issue/IDEA-3...

Really like the templating capability of last #maven dependency plugin, so easy to generate #asciidoc from your deps now!

living #doc with #maven is one more step easier

logging API should expose file or class name + method name at least · Issue #110737 · dotnet/runtime Logger frameworks log the source of the log message. In can be (file+method) or (class+method) - optionally with line number - but this is key to have some usefulness of the logs until you do it ma...

And another one

> github.com/dotnet/runti...

Who care about class or method names in logs?....seriously

[API Proposal]: simple console logger should support singleline except for exception use case · Issue #110793 · dotnet/runtime Background and motivation Add a toggle (maybe an enum instead of current toggle which would be deprecated) to enable to use single line formatting except for the exception stacktrace which is alway...

#dotnet "common" logging scenario looks weird, the common need seems to not have anything usable when enabling single line

> github.com/dotnet/runti...

First JDK 26 Release Candidate: mail.openjdk.org/pipermail/jd...

Downloads: jdk.java.net/26/

#JDK26 #Java26 #OpenJDK #TestItNow

Well, best workaround is "-Dsurefire.enableProcessChecker=" or set it in the pom or leave it blank....this is actually no more needed since we have container based CI.

#dotnet tip:

> export MSBUILDTERMINALLOGGER=off

and you get back useful logs on dotnet commands

Testing your #java deployments (#bundlebee), upgrades, operator with #junit never had been so easy!

Fun fact, how to go faster than HTTP/2.0 on an API?
....
Using HTTP/1.1 :D

The story behind: being able to use the MAX_CONCURRENT_STREAMS value to respect it can be very challenging depending the lib you do use so 1.1 *can* be faster for you

� Allez tous vous faire foutre ! � Rob Pike explose contre la GenAI : quand l'un des cr�ateurs de Go~? Plan 9~? UTF-8 accuse l'IA de piller~? polluer et d�truire le sens m�me du progr�s technologique Rarement une figure fondatrice de l�informatique moderne aura exprim� une col�re aussi brute, aussi personnelle, aussi d�sabus�e. En quelques messages, Rob Pike a fait exploser le vernis polic� du d�b...

Ce coup de gueule de Rob Pike (co-créateur de Go, et de l’UTF-8) sur l’IAGen est salutaire et rééquilibre l’ambiance positiviste et dépourvue de recul qui domine dans la communauté IT.
La dissonance cognitive n’est jamais confortable mais elle est souvent salvatrice…

yep (and fail the build if requested), guess you can write a jbang command to make it or workaround it with ejecting as a maven project (not sure it would be good but that's how I see it can work today)

say me ossindex plugin too and I can test it again (seriously)

I can but I literally don't get benefits but only drawbacks.
I can accept a few drawbacks if it becomes a one trivial command (x script.java) else I'd just go with #maven and keep the sane existing and shared ecosystem about vulnerability scanning, automatic upgrades etc...

maven is supported everywhere, trivial to setup
jbang requires some knowledge, plugins not working always very well (vscode integration is not great - not jbang plugin fault only, eclipse stack is to blame also) and is a new toy not that well known
so => it is negative since it doesnt make it better

FTR, being maven (or resolver) first is my current _default_ option and it works super well with plain java so jbang gain would be negative if I go that path

IDE is 80% about completion, if it can run it is great, if not console is fine I think. Challenge there is really completion only.
And if I need a whatever.build.file I'd just make it a pom.xml and be it, will work and be integrated everywhere.

Guess it will be only usable better once openjdk.org/jeps/468 pops up by default, but we're on track!

Yes, and do not get it wrong, it fails as the expectation of jbang to be better than literally "mvn" command in terms of work for the end users.
We're in IT, everything always works ;), it is just not as frictless as I expected there "everywhere"

I'd say it works for your case, for mine it breaks it and the gain I was hoping to move from either maven or 2 different commands to jbang is actually not there, so it doesn't for me but it is ok ;)

ack that, was trying to get a one liner to get started for non "native java" guys but using stdout can not be the best option even if the most convenient (in particular with the new IO class).

(following of previous answer - NotEnoughCharException ;)) issue is on env where you also assert stderr is clean for ran scripts only - and not only exit code. Same spirit than the old jenkins rule "no warning at all" if you recall this plugin.

not much a preference thing, it is literally that output must be yaml/json to be piped so it you start with logger lines you break it. Since quiet can't be in a directive it ends as being the script hacked on the shebang line or another script/file which defeats the whole jbang usage there

Security -> doesn't change anything until you forbid remotting by default which would bring security, logging doesn't make it smoother strictly speaking.

gradle disables info by def by ex

Use case -> www.yupiik.io/kubernetes-j... (println is broken by dep logs - it is fine to log errors, not info)

sadly none of these do make thing better than current state (a small pom):
1) quiet will not magically be set on all machines (directive?) so need another script/file anyway
2) exporting it is like not using jbang (works well, just too much for a script of 50 lines IMHO)

played a lot with #java 25, #jbang (why not quiet by default!), #groovy #grab, #kotlin #dependson and none is that smooth when it comes to handle dependencies with real portable IDE support...think I'll go with #java 25 and a pom.xml to handle my scripts with deps for now

Did a PR to refesh @Yupiik #kubernetes #java #descriptors now we have #java 25 and unamed classes

> github.com/yupiik/kuber...
/
> github.com/yupiik/kuber...

#kubernetes #descriptor #generation #industrialization

And here it is, the #jsonschema visually browsable in a web UI for #kubernetes!

> www.yupiik.io/bundlebee/ge...

Really nicer than the plain #json ones in #k8s repository or REST API schemas!

#yupiik #easines #cloud #helm #bundlebee #kustomize #descriptors