Re: [ANN] Clojure 1.7.0-alpha5 now available

I added a comment in CLJ-979, I believe I have a case which is broken by the change in delegation order, if I'm correct in my understanding of what

Re: [ANN] Clojure 1.7.0-alpha5 now available

Thank you very much Nicola! On Thu, Jan 29, 2015 at 7:24 PM, Nicola Mometto wrote: > > Jozef, I looked into this and opened a ticket with a proposed patch to > fix this issue: http://dev.clojure.org/jira/browse/CLJ-1650 > > Jozef Wagner writes: > > > With CLJ-979 applied, lein may crash when doi

Re: [ANN] Clojure 1.7.0-alpha5 now available

Jozef, I looked into this and opened a ticket with a proposed patch to fix this issue: http://dev.clojure.org/jira/browse/CLJ-1650 Jozef Wagner writes: > With CLJ-979 applied, lein may crash when doing :aot :all. This is caused > when lein forces recompilation of already compiled classes. Not su

Re: [ANN] Clojure 1.7.0-alpha5 now available

With CLJ-979 applied, lein may crash when doing :aot :all. This is caused when lein forces recompilation of already compiled classes. Not sure whether this should be considered a lein of clojure bug though. See https://github.com/wagjo/aotbug for contrived example. Jozef On Saturday, January 1

Re: [ANN] Clojure 1.7.0-alpha5 now available

This is a different though related problem to http://dev.clojure.org/jira/browse/CLJ-1315 That problem was trying to solve "require" un-necessarily importing classes, your example is a type hint with a similar issue. I do think you're correct, it looks like a reasonable change, but it would requi

Re: [ANN] Clojure 1.7.0-alpha5 now available

Just to follow up, doing the obvious (though perhaps naïve) fix of changing forName to forNameNonLoading in the referenced maybeClass does seem to fix the issue and I can now AOT compile my application without needing the binaries for this machine. I don't know whether this change would have any ot

Re: [ANN] Clojure 1.7.0-alpha5 now available

I think I am seeing static initialisers being run at AOT time when the class is used as a type hint tag. This isn't a regression from previous versions of clojure, but with the forNameNonLoading changes that have gone in recently I was under the impression that this shouldn't be a problem any more.

Re: [ANN] Clojure 1.7.0-alpha5 now available

Sweet - thanks for the reply. I hope I didn't sound impatient... I'm very grateful for your hard work on it. I'd offer to help but I'm sure it is beyond me and my crazy-mad Clojure skillz(tm). I think I'll leave the hard stuff to the real experts :-) Alan On Wednesday, January 21, 2015 at 6:07

Re: [ANN] Clojure 1.7.0-alpha5 now available

I am actively working on an updated set of patches, hopefully for inclusion in the next alpha. Alex -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are mo

Re: [ANN] Clojure 1.7.0-alpha5 now available

How is the feature expressions feature progressing? I haven't heard much about it recently - we are looking at converting a CLJS project from using crossovers to feature expressions... Alan On Sunday, January 11, 2015 at 6:00:52 PM UTC-8, Alex Miller wrote: > > No, we do not work from planned r

Re: [ANN] Clojure 1.7.0-alpha5 now available

Just FYI, we took 1.7.0-alpha5 into production today. We’ll let you know if we run into any issues. Other than tripping over the AOT/JIT thing with core.typed, it’s been smooth sailing on dev/QA… …and we actually have a transducer in production now! Sean On Jan 10, 2015, at 7:36 AM, Alex Mille

Re: [ANN] Clojure 1.7.0-alpha5 now available

Just to follow up on this, Tom reported to me on Twitter that the box giving slow results was bad and the regression was not repeatable on another box. Alex On Monday, January 12, 2015 at 12:37:05 PM UTC-6, tcrayford wrote: > > I said this already on twitter, but I see heavy (5-10x slower) perf

Re: [ANN] Clojure 1.7.0-alpha5 now available

Alpha5 works for me. I noticed that the doc-string for *unchecked-math* doesn't mention the :warn-on-boxed behavior. Suggestion: "While bound to true, compilations of +, -, *, inc, dec and the coercions will be done without overflow checks. While bound to :warn-on-boxed, a warning will

Re: [ANN] Clojure 1.7.0-alpha5 now available

I discovered this problem with Cursive too. Stephen Gilardi had a great analysis when I asked about it here: https://groups.google.com/d/topic/clojure/wrVFuCjf0_Y/discussion. tl;dr - Clojure only produces an error on dependency cycles when the dependencies appear in the ns form, or at least did unt

Re: [ANN] Clojure 1.7.0-alpha5 now available

Nicola, who is patient enough to remind me of things we have talked about already before. Now I recall running across this weirdness in the Manifold library before, since it caused Eastwood to fail when run on Manifold. Created a Clojure ticket to record anything discovered about this issue, or a

Re: [ANN] Clojure 1.7.0-alpha5 now available

There actually is a dependency cycle in manifold, it's just not in the ns macro: manifold.stream.graph depends on manifold.graph here: https://github.com/ztellman/manifold/blob/master/src/manifold/stream/graph.clj#L5 maniphold.stream depends on manifold.stream.graph here: https://github.com/ztell

Re: [ANN] Clojure 1.7.0-alpha5 now available

I can reproduce this. I have also determined that the change in behavior is due to the patch applied for ticket CLJ-1544 [1] [2]. There doesn't appear to be an actual dependency cycle in the manifold lib, but I've only looked at this for a few mins. Andy [1] http://dev.clojure.org/jira/browse/C

Re: [ANN] Clojure 1.7.0-alpha5 now available

I noticed that with alpha5, I can no longer build an uberjar with {:aot :all} if I depend on [gloss "0.2.4"]: $ lein uberjar Compiling alpha5-test.core java.lang.Exception: Cyclic load dependency: [ /manifold/stream ]->/manifold/stream/graph->[ /manifold/stream ]->/byte_streams->/gloss/core/for

Re: [ANN] Clojure 1.7.0-alpha5 now available

Oh, and all our apps now pass / work correctly on 1.7.0-alpha5 (I stripped Typed Clojure out of our code base to make it all work). Sean > On Jan 12, 2015, at 6:25 PM, Sean Corfield wrote: > > On Jan 12, 2015, at 11:32 AM, Sean Corfield > wrote: >> http://dev.clojure

Re: [ANN] Clojure 1.7.0-alpha5 now available

On Jan 12, 2015, at 11:32 AM, Sean Corfield wrote: > http://dev.clojure.org/jira/browse/CLJ-1639 > It turned out to be due to core.typed whose JAR includes AOT’d versions of core.cache and core.memoize amongst others. I’ve left this issue open with

Re: [ANN] Clojure 1.7.0-alpha5 now available

http://dev.clojure.org/jira/browse/CLJ-1639 > On Jan 12, 2015, at 10:37 AM, Alex Miller wrote: > > I would be helpful to me at this point to have a jira regarding this problem > which I will presume for the time being is the same in both Andy and S

Re: [ANN] Clojure 1.7.0-alpha5 now available

I said this already on twitter, but I see heavy (5-10x slower) perf regressions on my benchmark suite from 1.7-alpha5 vs 1.6. I am trying the other alphas now - I have a comprehensive benchmark suite that takes 4 hours or so to run, and it's a bit backed up right now (I only have one box to run

Re: [ANN] Clojure 1.7.0-alpha5 now available

I would be helpful to me at this point to have a jira regarding this problem which I will presume for the time being is the same in both Andy and Sean's case. On Monday, January 12, 2015 at 11:30:57 AM UTC-6, Andy Fingerhut wrote: > > Nicola, Sean: > > I am _not_ testing with Sean's code, so can

Re: [ANN] Clojure 1.7.0-alpha5 now available

Nicola, Sean: I am _not_ testing with Sean's code, so can't comment whether my results will correspond with his. I have found that between Clojure 1.7.0-alpha4 and 1.7.0-alpha5, and the latest Eastwood version 0.2.1 released, I get an exception trying to lint the namespace clojure.reflect in Cloj

Re: [ANN] Clojure 1.7.0-alpha5 now available

I have confirmed that reverting this commit addresses the problems we are encountering: https://github.com/clojure/clojure/commit/9f277c80258b3d2951128ce26a07c30ad0b47af0 "CLJ-979: make clojure resolve to the correct Class instances" Since we routinely run our test suite against master, and we

Re: [ANN] Clojure 1.7.0-alpha5 now available

On Jan 12, 2015, at 8:09 AM, Nicola Mometto wrote: > Can you try a custom version of clojure 1.7.0-alpha5 with commit > e5a104e894ed82f244d69513918d570cee5df67d reverted and see if you still > get the exception? Yes, reverting that one commit and creating Clojure 1.7.0-aot5 and rebuilding and te

Re: [ANN] Clojure 1.7.0-alpha5 now available

Can you try a custom version of clojure 1.7.0-alpha5 with commit e5a104e894ed82f244d69513918d570cee5df67d reverted and see if you still get the exception? Nicola Sean Corfield writes: > Here’s what I see reproducing this in the REPL - this main project doesn't > use core.async (but does use a

Re: [ANN] Clojure 1.7.0-alpha5 now available

Here’s what I see reproducing this in the REPL - this main project doesn't use core.async (but does use a very small AOT'd library). I'm going to try to cut this down to see if I can repro outside our main project (worldsingles.cache is a thin wrapper over core.cache and core.memoize - moving th

Re: [ANN] Clojure 1.7.0-alpha5 now available

On Jan 12, 2015, at 3:00 AM, Bronsa wrote: > Do the failing projects require AOT compilation? we used to see a similar > exception in eastwood when reloading core.cache and one of the AOT patches > committed can cause some namespaces to be reloaded > We have one (very small) AOT’d library tha

Re: [ANN] Clojure 1.7.0-alpha5 now available

Yes, ArrayList is a particularly common case (see http://insideclojure.org/2015/01/07/vec-perf/). I'm going to reinstate the prior one though, no reason to break people (even if they are using an internal method). On Monday, January 12, 2015 at 3:59:09 AM UTC-6, Mikera wrote: > > It seems a lit

Re: [ANN] Clojure 1.7.0-alpha5 now available

Do you have the full stack trace in the cases where you see this without core.async? On Sunday, January 11, 2015 at 7:17:27 PM UTC-6, Sean Corfield wrote: > > I tried upgrading a few more apps and ran into this same problem in the > absence of core.async - but in the presence of core.cache and c

Re: [ANN] Clojure 1.7.0-alpha5 now available

Rangel - I have filed http://dev.clojure.org/jira/browse/CLJ-1638 to replace the method, which seems like the best solution. On Monday, January 12, 2015 at 3:43:09 AM UTC-6, Rangel Spasov wrote: > > Hey guys, > > I noticed that PersistentVector.create() is missing one arity that used to > exist

Re: [ANN] Clojure 1.7.0-alpha5 now available

I have a set of micro-benchmarks on a project that is compute-heavy (symbolic regression, genetic programming). I see no performance regressions :-) 1.6.0 http://viewer.gorilla-repl.org/view.html?source=gist&id=3e943292b03755681e50&filename=expression-1.6.0.clj 1.7.0-alpha5 http://viewer.gorill

Re: [ANN] Clojure 1.7.0-alpha5 now available

Do the failing projects require AOT compilation? we used to see a similar exception in eastwood when reloading core.cache and one of the AOT patches committed can cause some namespaces to be reloaded Il giorno 12/gen/2015 02:17, "Sean Corfield" ha scritto: > I tried upgrading a few more apps and

Re: [ANN] Clojure 1.7.0-alpha5 now available

It seems a little odd that there is an "ArrayList" override but not "List". Presumably that is there as an optimisation since ArrayList is such a common case? It seems to have happened in this commit for CLJ-1546, which replaces the "List" override with "ArrayList" https://github.com/clojure/cl

Re: [ANN] Clojure 1.7.0-alpha5 now available

Hey guys, I noticed that PersistentVector.create() is missing one arity that used to exist before: PersistentVector.create(List items) ... which caused this library https://github.com/ninjudd/clojure-protobuf/blob/develop/src/flatland/protobuf/PersistentProtocolBufferMap.java#L479 to throw

Re: [ANN] Clojure 1.7.0-alpha5 now available

Alex, 1.7.0-alpha5 looks OK to me. It passed with some regression testing for an AOTd http-kit compojure / core.async REST server using Lucene, clojure/java.jdbc / MySQL, Java 1.7.0_65 on Ubuntu 10.04.1 LTS. -- You received this message because you are subscribed to the Google Groups "Clojure"

Re: [ANN] Clojure 1.7.0-alpha5 now available

On Jan 11, 2015, at 6:13 PM, Alex Miller wrote: > Thanks Sean. I'm hoping to get new releases of core.cache and core.menoize in > th near future too but seems worth looking into. I'm having a hard time producing a standalone repro case. It fails in the context of every single one of our apps, b

Re: [ANN] Clojure 1.7.0-alpha5 now available

Thanks Sean. I'm hoping to get new releases of core.cache and core.menoize in th near future too but seems worth looking into. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that pos

Re: [ANN] Clojure 1.7.0-alpha5 now available

No, we do not work from planned release dates. When it's done, we ship it. The current plan is to do at least one more alpha with a version of feature expressions that is a bit slimmer than the current available patches to make that available for people to experiment (presumably along with tool

Re: [ANN] Clojure 1.7.0-alpha5 now available

I tried upgrading a few more apps and ran into this same problem in the absence of core.async - but in the presence of core.cache and core.memoize so I'm trying to create a small test case to isolate the issue. Sean > On Jan 11, 2015, at 5:07 PM, Sean Corfield wrote: > > The full Caused by st

Re: [ANN] Clojure 1.7.0-alpha5 now available

The full Caused by stacktrace is this: Caused by: java.lang.IllegalArgumentException: No implementation of method: :has? of protocol: #'clojure.core.cache/CacheProtocol found for class: clojure.core.memoize.PluggableMemoization at clojure.core$_cache_protocol_fn.invoke(core_deftype.clj:5

Re: [ANN] Clojure 1.7.0-alpha5 now available

On Jan 11, 2015, at 6:34 AM, Alex Miller wrote: > I would greatly appreciate hearing any feedback about this (or any other) > alpha, even if it's just: everything looks ok. I upgraded an app that uses core.async (0.1.346.0-17112a-alpha) and got this exception on the ns that brings in clojure.c

Re: [ANN] Clojure 1.7.0-alpha5 now available

So far so good, will go in prod with it next Wednesday. Will run some heavy integrated tests in the next 48 hours. Will report if anything shows up. Thank you, Luc P. > For my projects swapping 1.7.0-alpha4 -> alpha5 has not culminated in any > abnormalities. So... looking good thus far on my e

Re: [ANN] Clojure 1.7.0-alpha5 now available

For my projects swapping 1.7.0-alpha4 -> alpha5 has not culminated in any abnormalities. So... looking good thus far on my end. On Sunday, January 11, 2015 at 11:45:50 AM UTC-5, Plinio Balduino wrote: > > Hi there, Alex and Clojure team > > Is there a planned date for the stable version release?

Re: [ANN] Clojure 1.7.0-alpha5 now available

Hi there, Alex and Clojure team Is there a planned date for the stable version release? Regards Plínio On Sun, Jan 11, 2015 at 12:34 PM, Alex Miller wrote: > I would greatly appreciate hearing any feedback about this (or any other) > alpha, even if it's just: everything looks ok. > > We've ha

[ANN] Clojure 1.7.0-alpha5 now available

I would greatly appreciate hearing any feedback about this (or any other) alpha, even if it's just: everything looks ok. We've had a couple of regressions reported and that is hugely helpful as we can quickly turn around fixes for the next one. Interested particularly in: regressions, perfor

[ANN] Clojure 1.7.0-alpha5 now available

Clojure 1.7.0-alpha5 is now available. Try it via - Download: https://repo1.maven.org/maven2/org/clojure/clojure/1.7.0-alpha5/ - Leiningen: [org.clojure/clojure "1.7.0-alpha5"] A few of the highlights in alpha5: 1) New transducer arities for map-indexed, distinct, and interpose. The transducer v