Version 0.16 removes a symbol and also breaks njs

[kapouer]

Hi,

I'm not sure if it's in 0.16.1 with 6a2c5f3 or already was in 0.16.0 (my 0.16.0 wasn't building) but these changes happened between 0.15.0 and 0.16.1:

--- a/debian/libqjs0.symbols
+++ b/debian/libqjs0.symbols
@@ -175,8 +175,10 @@ libqjs.so.0 libqjs0 #MINVER#
  JS_NewUint8ArrayCopy@Base 0.13.0
  JS_ParseJSON@Base 0.13.0
  JS_PreventExtensions@Base 0.13.0
+ JS_PromiseMarkAsHandled@Base 0.16.1
  JS_PromiseResult@Base 0.13.0
  JS_PromiseState@Base 0.13.0
+ JS_PromiseThen@Base 0.16.1
  JS_ReadObject2@Base 0.13.0
  JS_ReadObject@Base 0.13.0
  JS_ResetUncatchableError@Base 0.13.0
@@ -261,7 +263,7 @@ libqjs.so.0 libqjs0 #MINVER#
  js_module_load@Base 0.13.0
  js_module_loader@Base 0.13.0
  js_module_set_import_meta@Base 0.13.0
- js_realloc2@Base 0.13.0
+#MISSING: 0.16.1-1# js_realloc2@Base 0.13.0
  js_realloc@Base 0.13.0
  js_realloc_rt@Base 0.13.0
  js_std_add_helpers@Base 0.13.0

Also, njs is broken by the same commit, see nginx/njs#1113.

It makes sense (after all, the 0.15 to 0.16 version bump follows semantic versioning), yet I wonder if this was intentional.

[saghul]

Yes those changes are intentional.

I removed realloc2 because it never quite had a big impact, and that is specially more true with the arena allocator.

[kapouer]

Thank you for your reply, closing, then.

[bnoordhuis]

I've opened nginx/njs#1114 with some fixes. njs works for me locally again when I include #1670.

@saghul wdyt about adding nginx/njs master to our CI somehow? Breaking downstream doesn't have to be a blocker but at least it's nice if we know and can put out advance warning.

[saghul]

Sounds good 👍

I generally know because of Txiki.js but the api surface is big so yeah I see how it can help!

[bnoordhuis]

What's the best way to set it up? Is there a way to mark a failing buildbot as informative rather than full-blown error? Far as I can tell that's a six-years-old-and-still-open feature request: actions/runner#2347

[kapouer]

In that case I would do like coverage, append the result to GITHUB_STEP_SUMMARY.

[bnoordhuis]

That only shows up on the Actions summary tab though, right? I rarely look at that. Ideally it shows up on the PR itself.

[saghul]

Maybe the CInscript can catch the failure, post a comment and exit successfully?

[bnoordhuis]

If you know how, I'll be happy to review the PR :-)

[saghul]

😅 I was thinking out loud...