Command box is using the 5.3 beta?

just noticed the docs ci process is using the 5.3 beta via command box, is this intended @bdw429s ?

https://travis-ci.org/lucee/lucee-docs/builds/379987506

That’s a server starting, not the core cli. I added the 5.3 cf engine to forgebox today, but it shouldn’t get pulled by default since it’s a beta. That might be a 3.9 behavior. I’ll check it tonight.

ah well, all the new 5.3 alias functions just got automatically published to docs :slight_smile:

i.e

Maybe we should be branching docs by version number or something, but it just adds the complexity of an already difficult job.

At the current rate of release 5.3 won’t be official for 2 more months; ie next RC, then GA.

at least the local bundled docs always match the current version :slight_smile:

there is a Introduced attribute which gets displayed but it’s doesn’t seem to be set all that often, the only ones I’ve seen are for 5.0.0.0

more importantly, we don’t have a process to sync edits in the docs back to the original xml definition.

the docs only import first time it encounters a function, tag, attribute or argument.

Any subsequent edits to the source xml file above don’t get imported as it might overwrite edits (but only if the generated .md file is committed, the publish process auto generates the docs, but it doesn’t commit them back to git)

Ok, I figured this out. Turns out it was my fault! When I added the 5.3 beta to ForgeBox today I failed to follow the proper npm-style semver and I called the version 5.3.0+86-beta (I turn the 4th number into a build number) instead of 5.3.0-beta+86 (Prerelease IDs come before build numbers). As such, the beta was registering as a “stable” version and was picked up as the “latest stable”. A good reason to dial in an exact version on your production CommandBox servers :slight_smile: I unpublished the incorrect version and re-published the 5.3 beta using the correctly formatted semver. If you re-run the Travis build, it should start up a 5.2.7 server now as the latest stable. Sorry again for mixing that up. I tested the 5.3 version to make sure it worked, but hadn’t tested a lucee@5 start!

I re-ran the Travis build and it’s back to 5.2.7. The question now is whether the docs are smart enough to remove functions that aren’t there any longer :slight_smile:

as the 5.3 changes weren’t committed to git, it just depends if the s3 website sync removes files which don’t exist any more, which i think it does?