Re: [racket-dev] What's the process for adding a project to DrDr
DrDr has the capability of doing that [installing packages], but installing Planet packages requires running raco setup which could break the parallelism and cause changes that are unsafe to the other tests. Also, I'd have to change it to go run all the files inside the package, which is do-able. On Tue, Oct 23, 2012 at 9:44 PM, Robby Findler wrote: > Would it make sense to have drdr get whalesong directly from planet? > > Robby > > On Mon, Oct 22, 2012 at 8:17 PM, Jay McCarthy wrote: >> We haven't done it before yet. >> >> The simplest way is to commit something to the core, but that's not >> good in the long run. >> >> One thing that we could do is create a directory like >> >> meta/drdr/external >> >> of scripts that download something from the Internet and run it. We'll >> have to think of a good way to make it secure (such as having this >> script know the sha1 of the zip it is downloading.) Also, for >> parallelization purposes, it will be important for it not to require >> the use of "raco setup" to get started. >> >> Does that sound plausible for Whalesong? If so, I can work out a >> template for you. >> >> Jay >> >> On Mon, Oct 22, 2012 at 5:51 PM, Danny Yoo wrote: >>> During release testing for 5.3.1, I found that Whalesong broke in a >>> few ways. I'd like to detect the problem in a better way than what >>> I've been doing now. What's involved in getting into the DrDr >>> automatic build bot, and how would I get it to track Whalesong as I >>> continue to develop it? >>> _ >>> Racket Developers list: >>> http://lists.racket-lang.org/dev >> >> >> >> -- >> Jay McCarthy >> Assistant Professor / Brigham Young University >> http://faculty.cs.byu.edu/~jay >> >> "The glory of God is Intelligence" - D&C 93 >> _ >> Racket Developers list: >> http://lists.racket-lang.org/dev -- Jay McCarthy Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay "The glory of God is Intelligence" - D&C 93 _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] What's the process for adding a project to DrDr
Would it make sense to have drdr get whalesong directly from planet? Robby On Mon, Oct 22, 2012 at 8:17 PM, Jay McCarthy wrote: > We haven't done it before yet. > > The simplest way is to commit something to the core, but that's not > good in the long run. > > One thing that we could do is create a directory like > > meta/drdr/external > > of scripts that download something from the Internet and run it. We'll > have to think of a good way to make it secure (such as having this > script know the sha1 of the zip it is downloading.) Also, for > parallelization purposes, it will be important for it not to require > the use of "raco setup" to get started. > > Does that sound plausible for Whalesong? If so, I can work out a > template for you. > > Jay > > On Mon, Oct 22, 2012 at 5:51 PM, Danny Yoo wrote: >> During release testing for 5.3.1, I found that Whalesong broke in a >> few ways. I'd like to detect the problem in a better way than what >> I've been doing now. What's involved in getting into the DrDr >> automatic build bot, and how would I get it to track Whalesong as I >> continue to develop it? >> _ >> Racket Developers list: >> http://lists.racket-lang.org/dev > > > > -- > Jay McCarthy > Assistant Professor / Brigham Young University > http://faculty.cs.byu.edu/~jay > > "The glory of God is Intelligence" - D&C 93 > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] chaperones in 5.3.1?
Is it possible that one of those vectors is mutable and the other is immutable? Robby On Tue, Oct 23, 2012 at 7:03 PM, Danny Yoo wrote: > I'm seeing the following error deep within Whalesong when I'm running > its test suite: > > > . . > ../../../Library/Racket/planet/300/5.3.0.900/cache/dyoo/browser-evaluate.plt/1/0/browser-evaluate.rkt:145:0: > prop:chaperone-contract: expected a chaperone of > '#("#" > 401 14 10 9), got > '#("#" > 401 14 10 9) > > > I'm confused (as usual), because my browser-evaluate library shouldn't > be doing anything to chaperoned values. The error makes it appear > that some value got incorrectly unwrapped somehow. I'll try to trace > this down better, but has anyone else seen this before? > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] Is this a type error?
>> >> (: current-defined-name (Parameterof (U Symbol False LamPositionalName))) >> (define current-defined-name (make-parameter UNKNOWN)) > > > That looks to me like a problem with the inferencer. There's some interaction with the structure in the union. If I simplify the type definition to: (: current-defined-name (Parameterof (U Symbol))) then the type system seems to be happy to accept 'unknown as an acceptable symbol. _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] Is this a type error?
On 10/23/2012 05:19 PM, Danny Yoo wrote: #lang typed/racket/base ;; We may have more information about the lambda's name. This will show it. (define-struct: LamPositionalName ([name : Symbol] [path : String];; the source of the name [line : Natural] [column : Natural] [offset : Natural] [span : Natural]) #:transparent) (: current-defined-name (Parameterof (U Symbol LamPositionalName))) (define current-defined-name (make-parameter 'unknown)) To add, I can work around this behavior by doing: (: UNKNOWN Symbol) (define UNKNOWN 'unknown) (: current-defined-name (Parameterof (U Symbol False LamPositionalName))) (define current-defined-name (make-parameter UNKNOWN)) That looks to me like a problem with the inferencer. It's *something* with the inferencer, because you can work around it by annotating: (: current-defined-name (Parameterof (U Symbol LamPositionalName))) (define current-defined-name (make-parameter (ann 'unknown Symbol))) I just don't know whether it's technically an error in TR. Neil ⊥ _ Racket Developers list: http://lists.racket-lang.org/dev
[racket-dev] chaperones in 5.3.1?
I'm seeing the following error deep within Whalesong when I'm running its test suite: . . ../../../Library/Racket/planet/300/5.3.0.900/cache/dyoo/browser-evaluate.plt/1/0/browser-evaluate.rkt:145:0: prop:chaperone-contract: expected a chaperone of '#("#" 401 14 10 9), got '#("#" 401 14 10 9) I'm confused (as usual), because my browser-evaluate library shouldn't be doing anything to chaperoned values. The error makes it appear that some value got incorrectly unwrapped somehow. I'll try to trace this down better, but has anyone else seen this before? _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] Is this a type error?
On Tue, Oct 23, 2012 at 5:14 PM, Danny Yoo wrote: > The following program is generating a type error in 5.3.1, but I don't > understand it yet. > > ;;; > #lang typed/racket/base > > ;; We may have more information about the lambda's name. This will show it. > (define-struct: LamPositionalName ([name : Symbol] >[path : String];; the source of the > name >[line : Natural] >[column : Natural] >[offset : Natural] >[span : Natural]) #:transparent) > > > (: current-defined-name (Parameterof (U Symbol LamPositionalName))) > (define current-defined-name (make-parameter 'unknown)) To add, I can work around this behavior by doing: (: UNKNOWN Symbol) (define UNKNOWN 'unknown) (: current-defined-name (Parameterof (U Symbol False LamPositionalName))) (define current-defined-name (make-parameter UNKNOWN)) but I'd still like to know why the earlier code doesn't work in 5.3.1. _ Racket Developers list: http://lists.racket-lang.org/dev
[racket-dev] Is this a type error?
The following program is generating a type error in 5.3.1, but I don't understand it yet. ;;; #lang typed/racket/base ;; We may have more information about the lambda's name. This will show it. (define-struct: LamPositionalName ([name : Symbol] [path : String];; the source of the name [line : Natural] [column : Natural] [offset : Natural] [span : Natural]) #:transparent) (: current-defined-name (Parameterof (U Symbol LamPositionalName))) (define current-defined-name (make-parameter 'unknown)) ;;; I see the following error message: . Type Checker: Expected (Parameterof (U Symbol LamPositionalName)), but got (Parameterof (U Symbol LamPositionalName 'unknown)) in: (make-parameter (quote unknown)) which I didn't expect because 'unknown should be a symbol. _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] bug: internal typechecker error
On 22/10/12 21:43, Paulo J. Matos wrote: Hello, As I was converting a program into typed racket in drscheme 5.3 I got an internal typechecker error. Load the attached file and press run. That should allow you to reproduce the problem. git HEAD has the problem fixed. -- PMatos _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] GC segfault
This fixes the problem. You are magic. -Ian - Original Message - From: "Matthew Flatt" To: "J. Ian Johnson" Cc: "dev" Sent: Tuesday, October 23, 2012 2:44:35 PM GMT -05:00 US/Canada Eastern Subject: Re: [racket-dev] GC segfault Thanks for the example! I've pushed a repair for a recent JIT change that inlines structure allocation. Let me know if it doesn't solve the problem. At Tue, 23 Oct 2012 07:30:20 -0700, Matthew Flatt wrote: > Smaller examples are better, but anything that I can run may be > helpful. > > At Tue, 23 Oct 2012 10:08:56 -0400 (EDT), "J. Ian Johnson" wrote: > > Every time I run my newly finished analysis on a benchmark, it crashes with > > > > Seg fault (internal error during gc) at 0x7fce50d3d010 > > SIGSEGV SEGV_ACCERR SI_CODE 2 fault on 0x7fce50d3d010 > > Aborted (core dumped) > > > > What extra info can I give to help diagnose the problem? The full source > > code > > is not likely to be illuminating. > > (tried in 5.3.1.1 and a pull from 15 minutes ago) > > -Ian > > _ > > Racket Developers list: > > http://lists.racket-lang.org/dev > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] GC segfault
Thanks for the example! I've pushed a repair for a recent JIT change that inlines structure allocation. Let me know if it doesn't solve the problem. At Tue, 23 Oct 2012 07:30:20 -0700, Matthew Flatt wrote: > Smaller examples are better, but anything that I can run may be > helpful. > > At Tue, 23 Oct 2012 10:08:56 -0400 (EDT), "J. Ian Johnson" wrote: > > Every time I run my newly finished analysis on a benchmark, it crashes with > > > > Seg fault (internal error during gc) at 0x7fce50d3d010 > > SIGSEGV SEGV_ACCERR SI_CODE 2 fault on 0x7fce50d3d010 > > Aborted (core dumped) > > > > What extra info can I give to help diagnose the problem? The full source > > code > > is not likely to be illuminating. > > (tried in 5.3.1.1 and a pull from 15 minutes ago) > > -Ian > > _ > > Racket Developers list: > > http://lists.racket-lang.org/dev > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] Pre-Release Checklist for v5.3.1, Second Call
Danny: I'm thinking that we should probably leave this in the checklist, like other important things that are tested automatically. Running the tests "by hand" once at release time seems just like an extra, comforting step. Robby On Mon, Oct 22, 2012 at 5:21 PM, Danny Yoo wrote: >> * Danny Yoo >> - Whalesong Tests (confirm that Whalesong runs from PLaneT) > > Does not run, but can you take this out of the checklist anyway? > > I'll follow up with a request on how to add Whalesong as a DrDr test > so that it's continually tested as opposed to tested only at > release-branch time. > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
Re: [racket-dev] GC segfault
Smaller examples are better, but anything that I can run may be helpful. At Tue, 23 Oct 2012 10:08:56 -0400 (EDT), "J. Ian Johnson" wrote: > Every time I run my newly finished analysis on a benchmark, it crashes with > > Seg fault (internal error during gc) at 0x7fce50d3d010 > SIGSEGV SEGV_ACCERR SI_CODE 2 fault on 0x7fce50d3d010 > Aborted (core dumped) > > What extra info can I give to help diagnose the problem? The full source code > is not likely to be illuminating. > (tried in 5.3.1.1 and a pull from 15 minutes ago) > -Ian > _ > Racket Developers list: > http://lists.racket-lang.org/dev _ Racket Developers list: http://lists.racket-lang.org/dev
[racket-dev] GC segfault
Every time I run my newly finished analysis on a benchmark, it crashes with Seg fault (internal error during gc) at 0x7fce50d3d010 SIGSEGV SEGV_ACCERR SI_CODE 2 fault on 0x7fce50d3d010 Aborted (core dumped) What extra info can I give to help diagnose the problem? The full source code is not likely to be illuminating. (tried in 5.3.1.1 and a pull from 15 minutes ago) -Ian _ Racket Developers list: http://lists.racket-lang.org/dev