Blocking bugs that should (?) go in before 1.5 release

[Andy Fingerhut]

Stuart Halloway asked for people to yell loudly on the dev group if there were bugs that should definitely go in before the next major release. Stuart Sierra mentioned recently that the plan is to go into beta for 1.5 soon. So far the only blocking bug anyone has mentioned is:

CLJ-1066 0001-Don-t-AOT-compile-clojure.core.reducers.patch

discussed here in mid September:

https://groups.google.com/forum/?fromgroups=#!topic/clojure-dev/PGKktFIXIts

I've included it on the prescreened patch list since then, near the beginning under a heading "Blocking bugs", but given the possibility of going to 1.5 beta soon I thought I'd call it out separately from the big prescreened patch list. If there are others you think ought to be on that list that block the 1.5 release, now might be a good time to mention them.

Andy

[Steve Miner]

CLJ-927 default tagged literal reader

There's no patch yet, but I will submit one tomorrow unless someone beats me to it.

[Tassilo Horn]

Andy Fingerhut <andy.fi...@gmail.com> writes:

> Stuart Halloway asked for people to yell loudly on the dev group if
> there were bugs that should definitely go in before the next major
> release. Stuart Sierra mentioned recently that the plan is to go into
> beta for 1.5 soon. So far the only blocking bug anyone has mentioned
> is:
>
> CLJ-1066 0001-Don-t-AOT-compile-clojure.core.reducers.patch

Yep, that's really a blocker.

The following aren't blockers, but they are simple and I see no reason
to defer them.

This one is pending, probably because Aaron Bedra added a comment that
isn't addressed until now, simply because I don't understand it. I
tried to contact Aaron per mail twice over the last weeks, but I didn't
get a response so far. But for me, the patch looks good and makes sense
as it is.

CLJ-1024 Varargs protococol impls can be defined but not called

This one is obvious, I don't see a reason to defer it.

CLJ-1084 (object-array [1]) is ~3x slower than (object-array (rseq [1]))

This doc-fix is pending, cause Rich hasn't responded if definterface is
meant to be public API.

CLJ-917 clojure.core/definterface is not included in the API docs

Bye,
Tassilo

[Chas Emerick]

I was planning on flogging these for Friday, but now's as good a time as any.

clojure.main/repl unconditionally refers REPL utilities into *ns*
http://dev.clojure.org/jira/browse/CLJ-1085

Don't squash explicit :line and :column metadata in the MetaReader
http://dev.clojure.org/jira/browse/CLJ-1079

Both have tested, "smallest-change" patches attached that apply cleanly. The former is particularly important to nREPL; the latter pairs well with the recent :column metadata enhancement, and makes using code transformation tools like kibit and cljx far more pleasant (i.e. with the patch, you won't lose line numbers for debugging purposes for transformed code).

Here's hoping :-),

- Chas

[Bronsa]

FWIW this is an incredibly trivial defect and the patch just adds a `declare`, it'd be annoying not to have this on 1.5

2012/10/17 Chas Emerick <ch...@cemerick.com>

--
You received this message because you are subscribed to the Google Groups "Clojure Dev" group.
To post to this group, send email to cloju...@googlegroups.com.
To unsubscribe from this group, send email to clojure-dev...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/clojure-dev?hl=en.

[Stuart Halloway]

I am working down this thread screening tickets, which means I am not
writing tests for http://dev.clojure.org/jira/browse/CLJ-960, which
Rich would very much like to have.

Anybody want to write some tests for CLJ-960?

[Chas Emerick]

There's some tests along these lines in the patch on CLJ-1079, though those use explicit :line and :column metadata.

I can add a patch to CLJ-960 with some tests along the same lines, plus walking through nested forms and such to verify their :column meta.

- Chas

[Chas Emerick]

Test patch now attached to CLJ-960.

- Chas

On Oct 19, 2012, at 11:25 AM, Stuart Halloway wrote:

[Jim Blomo]

http://dev.clojure.org/jira/browse/CLJ-862 fixes the interaction
between lazy pmap and chunked sequences. This is a pretty small
change, and having better defined behavior in a commonly used method
seems like a good candidate for 1.5.