Fork me on GitHub
#cljs-dev
<
2021-05-14
>
dnolen16:05:22

one more change for code splitting guide

dnolen16:05:11

looking at bumping Closure to prep for next release

dnolen16:05:19

if there's anything important you all think I should look at now is the time

henryw37408:05:17

Any ticket? https://clojure.atlassian.net/browse/CLJS-3294 would be great to fix IMO. Clearly a bunch of effort has gone in to data_literals.cljc already and hopefully not much more would be required to make it work

dnolen16:05:46

@alexmiller building google-closure-library would need a release of that as well

dnolen16:05:53

(unless that's automated now)

dnolen16:05:39

@alexmiller re: site needs another bump -- missed clj --main ...

Alex Miller (Clojure team)16:05:05

will build google-closure

dnolen16:05:08

I triggered the build, just needs release steps - thanks!

Alex Miller (Clojure team)16:05:43

yeah, I missed that and redid it :)

Alex Miller (Clojure team)16:05:59

I just released though, will take 10-15m to show up

Alex Miller (Clojure team)17:05:58

org.clojure/google-closure-library 0.0-20210512-04d6fb3b org.clojure/google-closure-library-third-party 0.0-20210512-04d6fb3b

dnolen17:05:10

@alexmiller looks like Google Closure Lib is now broken for how we consume it - a battle for another day

dnolen17:05:52

The stability of Closure Compiler and Lib has really fallen low

dpsutton17:05:00

do you mean it's buggy in the manner used in cljs or conceptually is moving away from the way cljs uses it?

dnolen17:05:22

No just that everytime I’ve bumped in the past two years stuff is broken

dnolen17:05:45

They don’t use semantic versioning but they also change stuff

dnolen17:05:27

We’re somewhat unique consumers but I’m not even so sure about that

dnolen17:05:58

My impression was that other users of Closure depend on some details as well for custom building

dnolen17:05:41

glib also has a lot of low value refactoring

dnolen17:05:09

Or perhaps can’t see how it could matter

dnolen17:05:49

Switching to ES6 classes and goog modules before

dnolen17:05:28

@alexmiller that's interesting but probably incomplete? If we could get them in JIRA could help them take that past the finish line

Alex Miller (Clojure team)18:05:19

I can fix them up, sounds like they have a CA already

Alex Miller (Clojure team)18:05:36

(although I always wonder why people have more than 20 parameters :)

borkdude18:05:10

perhaps a macro generated it?

Alex Miller (Clojure team)18:05:19

"If you have a procedure with 10 parameters, you probably missed some." - Alan Perlis

😆 8
dnolen19:05:18

I think bumping glib for next release is probably more trouble than it's worth - who knows what lurks beneath - can save it for later

dnolen19:05:32

Again feel free to propose other tickets worth considering

dnolen19:05:54

@alexmiller Windows support for clojure tool is pretty good now yes?

borkdude19:05:54

The main problems with it are: it's powershell, so shelling out to it from, say, Java, or nodeJS, or whatever editor is problematic, as it's not a normal process you can start. Another problem is that it doesn't work from cmd.exe, as you always need to go through powershell first.

borkdude19:05:59

https://github.com/borkdude/deps.clj solves some of these (in fairness, niche, because Windows usage is already niche in Clojure land) problems by providing a binary* instead which is a drop-in replacement for the official Clojure CLI *(or uberjar, or babashka script)

borkdude19:05:57

@pez can probably tell you more from the editor integration side, as he dealt with a fair share of Windows people lately

borkdude19:05:37

In general the -X style invocation causes some headaches for Windows users especially because quoting strings is pretty confusing in both cmd.exe and powershell

borkdude19:05:20

There's also #clj-on-windows if you want to know more.

Alex Miller (Clojure team)19:05:54

I think that is a fair accounting

borkdude19:05:18

Probably for the most mainstream usage it works well, this must also be said. People for whom it works well, are the ones you never hear from ;)

lread20:05:22

@dnolen on the Windows topic: way back I worked making https://clojure.atlassian.net/browse/CLJS-3098. I think you safely close this issue as stale. Would probably suggest rewriting bash scripts in Clojure if revisiting - and run through babashka or clojure.

pez20:05:06

Another problem on Windows is that the java.exe that Oracle’s installer makes the default, borks double quoting (strips double quotes fully and totally) . So any invokation of clojure using that java.exe, involving double quoting anything, will fail. This since JDK11. See https://bugs.java.com/bugdatabase/view_bug.do?bug_id=JDK-8266473

pez20:05:51

This is regardless of powershell or cmd.exe and hits deps.clj as well.

lread21:05:08

@dnolen If you are house-cleaning old issues, feel free to closeout https://clojure.atlassian.net/browse/CLJS-3075 too (kinda tied to Windows bash script work).

lread21:05:46

@dnolen unrelated to Windows, but while I’m looking at CLJS issues I worked on… I think you aren’t even building on Travis CI anymore? So https://clojure.atlassian.net/browse/CLJS-3114 can die too. Oh and https://clojure.atlassian.net/browse/CLJS-3079 seems to have dependencies on Windows issue, so feel free to kill it too.

henryw37408:05:17

Any ticket? https://clojure.atlassian.net/browse/CLJS-3294 would be great to fix IMO. Clearly a bunch of effort has gone in to data_literals.cljc already and hopefully not much more would be required to make it work