[bug] intmath.st check fails

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
19 messages Options
Reply | Threaded
Open this post in threaded view
|

[bug] intmath.st check fails

Stephen-71
Issue status update for
http://smalltalk.gnu.org/node/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   swoolerton
 Status:       active

Tested on Mac OS X 10.5.5 and RedHat Linux 5.2 64 bit.

When make check is run, test #7, intmath.st does not complete - the
test hangs. Comment added as below allows the test to complete:-

Eval [
   g := PiSeries new.
   9 timesRepeat: [ g := g accelerate ].
"    pi := g peek.
   pi numerator size printNl.
   pi denominator size printNl.
   (pi * 100000) asInteger "
]

The failure of intmath.st to complete has been observed on installs in
the 3.0.x series as well (Mac OS X, Redhat Linux and Gentoo Linux).

Note: When the tests were run in version 3.0.x, it was possible to
Ctrl-C when intmath.st hung, the testsuit would then continue. Under
3.1, pressing Ctrl-C does not halt the test (on Linux or Mac OS X), and
one has to go into the process list and kill the process, at which point
the tests will continue. The old behaviour was easier to work with...
comments welcome on what has changed and if we can have the old
behaviour back.

Thanks




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Stephen Woolerton

S.J.Chun wrote:
> For, if I built gst with gmp support, this problem does not appear.
>
Thanks for the note... given GMP is a numeric optimizer, I tried
reducing the number of iterations to see if it was just a timing thing
and got the following results (on OS X without GMP)

up to 7 iterations - pretty much straightaway
7 iterations  - 9 secs
8 iterations  - 2 minutes 50 secs
9 iterations  - approx 50 minutes

CPU: 2.4GHz Intel Core 2 Duo

So yes the test works.. no wonder I've never seen it finish until now!
Maybe that test should have a note by it to the effect that if you don't
have GMP loaded, forget it.

-------------------------
For context, the loop referred to is at the end of intMath.st and 9
iterations is what is shipped in the test ...

Eval [
     g := PiSeries new.
     9 timesRepeat: [ g := g accelerate ].
     pi := g peek.
     pi numerator size printNl.
     pi denominator size printNl.
     (pi * 100000) asInteger
]


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-2
Stephen Woolerton wrote:

>
> S.J.Chun wrote:
>> For, if I built gst with gmp support, this problem does not appear.
> Thanks for the note... given GMP is a numeric optimizer, I tried
> reducing the number of iterations to see if it was just a timing thing
> and got the following results (on OS X without GMP)
>
> up to 7 iterations - pretty much straightaway
> 7 iterations  - 9 secs
> 8 iterations  - 2 minutes 50 secs
> 9 iterations  - approx 50 minutes

Cool.  This has a been a longstanding issue, thanks for taking the time
to look through it.

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
 Status:       active

Fixed in 4c8038cb and backported to 3.0.x




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
-Status:       active
+Status:       fixed

forgot to change the status.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Giles Bowkett
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   gilesgoatboy
-Status:       fixed
+Status:       active

Hi, I changed the status to active because I still see this bug on OS X.
Someone who follows me on Twitter saw it too. We're both on OS X.
intmath.st fails, rockets to 99+% CPU, and hangs forever. make install,
and running GNU Smalltalk from the command line, succeed anyway.
MacPorts gives an error message if you try to install GNU Smalltalk via
the port package manager, saying it's unavailable for Intel Macs, which
in 2009 means most Macs, but in fact mine's Intel and it's running
fine. At least, it handles "Hello world!" and the interactive console.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Giles Bowkett
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   gilesgoatboy
 Status:       active

Version: we're both running the latest, 3.1.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Giles Bowkett
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   gilesgoatboy
 Status:       active

Also: it took a painful process of trial and error to determine how to
kill the failing test. My box was incapacitated for a while. Command-C
doesn't help. You have to check the processes and kill something
specific, I think gst.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: Re: [bug] intmath.st check fails

SeanTAllen
In reply to this post by Giles Bowkett

On Jan 17, 2009, at 5:33 PM, Giles Bowkett wrote:

> Issue status update for http://smalltalk.gnu.org/project/issue/268
> Post a follow up: http://smalltalk.gnu.org/project/comments/add/268
>
> Project:      GNU Smalltalk
> Version:      <none>
> Component:    Build
> Category:     bug reports
> Priority:     normal
> Assigned to:  Unassigned
> Reported by:  swoolerton
> Updated by:   gilesgoatboy
> -Status:       fixed
> +Status:       active
>
> Hi, I changed the status to active because I still see this bug on  
> OS X.
> Someone who follows me on Twitter saw it too. We're both on OS X.
> intmath.st fails, rockets to 99+% CPU, and hangs forever. make  
> install,
> and running GNU Smalltalk from the command line, succeed anyway.
> MacPorts gives an error message if you try to install GNU Smalltalk  
> via
> the port package manager, saying it's unavailable for Intel Macs,  
> which
> in 2009 means most Macs, but in fact mine's Intel and it's running
> fine. At least, it handles "Hello world!" and the interactive console.

Giles, I installed gst via MacPorts w/o an error message and have run  
both
that version and ones from built from source without issue. I'm using  
10.5.6.

I just ran intmath.st in less than 10 seconds with both the MacPorts  
version
and my built from source version.

Are you on something earlier than 10.5.6?




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: Re: [bug] intmath.st check fails

Giles Bowkett
Hi Sean, yes, I'm on 10.5.4. I pinged my Twitter friend to see if the
same's true for him. I'll try the system upgrade and see if that
helps.

On 1/17/09, Sean Allen <[hidden email]> wrote:

>
>  On Jan 17, 2009, at 5:33 PM, Giles Bowkett wrote:
>
>
> > Issue status update for
> http://smalltalk.gnu.org/project/issue/268
> > Post a follow up:
> http://smalltalk.gnu.org/project/comments/add/268
> >
> > Project:      GNU Smalltalk
> > Version:      <none>
> > Component:    Build
> > Category:     bug reports
> > Priority:     normal
> > Assigned to:  Unassigned
> > Reported by:  swoolerton
> > Updated by:   gilesgoatboy
> > -Status:       fixed
> > +Status:       active
> >
> > Hi, I changed the status to active because I still see this bug on OS X.
> > Someone who follows me on Twitter saw it too. We're both on OS X.
> > intmath.st fails, rockets to 99+% CPU, and hangs forever. make install,
> > and running GNU Smalltalk from the command line, succeed anyway.
> > MacPorts gives an error message if you try to install GNU Smalltalk via
> > the port package manager, saying it's unavailable for Intel Macs, which
> > in 2009 means most Macs, but in fact mine's Intel and it's running
> > fine. At least, it handles "Hello world!" and the interactive console.
> >
>
>  Giles, I installed gst via MacPorts w/o an error message and have run both
>  that version and ones from built from source without issue. I'm using
> 10.5.6.
>
>  I just ran intmath.st in less than 10 seconds with both the MacPorts
> version
>  and my built from source version.
>
>  Are you on something earlier than 10.5.6?
>
>
>


--
Giles Bowkett
http://gilesbowkett.com


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: Re: [bug] intmath.st check fails

SeanTAllen
I haven't used MacPorts much.. I basically started using it to get all  
the libraries I needed for gst..

Does it lock certain versions of libraries etc to OS versions?
Is it possible there is an issue with a version of a MacPorts library  
that gst uses?

On Jan 18, 2009, at 2:24 AM, Giles Bowkett wrote:

> Hi Sean, yes, I'm on 10.5.4. I pinged my Twitter friend to see if the
> same's true for him. I'll try the system upgrade and see if that
> helps.
>
> On 1/17/09, Sean Allen <[hidden email]> wrote:
>>
>> On Jan 17, 2009, at 5:33 PM, Giles Bowkett wrote:
>>
>>
>>> Issue status update for
>> http://smalltalk.gnu.org/project/issue/268
>>> Post a follow up:
>> http://smalltalk.gnu.org/project/comments/add/268
>>>
>>> Project:      GNU Smalltalk
>>> Version:      <none>
>>> Component:    Build
>>> Category:     bug reports
>>> Priority:     normal
>>> Assigned to:  Unassigned
>>> Reported by:  swoolerton
>>> Updated by:   gilesgoatboy
>>> -Status:       fixed
>>> +Status:       active
>>>
>>> Hi, I changed the status to active because I still see this bug on  
>>> OS X.
>>> Someone who follows me on Twitter saw it too. We're both on OS X.
>>> intmath.st fails, rockets to 99+% CPU, and hangs forever. make  
>>> install,
>>> and running GNU Smalltalk from the command line, succeed anyway.
>>> MacPorts gives an error message if you try to install GNU  
>>> Smalltalk via
>>> the port package manager, saying it's unavailable for Intel Macs,  
>>> which
>>> in 2009 means most Macs, but in fact mine's Intel and it's running
>>> fine. At least, it handles "Hello world!" and the interactive  
>>> console.
>>>
>>
>> Giles, I installed gst via MacPorts w/o an error message and have  
>> run both
>> that version and ones from built from source without issue. I'm using
>> 10.5.6.
>>
>> I just ran intmath.st in less than 10 seconds with both the MacPorts
>> version
>> and my built from source version.
>>
>> Are you on something earlier than 10.5.6?
>>
>>
>>
>
>
> --
> Giles Bowkett
> http://gilesbowkett.com



_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: Re: [bug] intmath.st check fails

Giles Bowkett
Hi Sean. I don't know, I don't use MacPorts much or know anything
about its internals. Anyway I checked with my Twitter friend and he's
on 10.5.6. So we're seeing the problem on 10.5.6 and 10.5.4.

On 1/18/09, Sean Allen <[hidden email]> wrote:

> I haven't used MacPorts much.. I basically started using it to get all the
> libraries I needed for gst..
>
>  Does it lock certain versions of libraries etc to OS versions?
>  Is it possible there is an issue with a version of a MacPorts library that
> gst uses?
>
>
>  On Jan 18, 2009, at 2:24 AM, Giles Bowkett wrote:
>
>
> > Hi Sean, yes, I'm on 10.5.4. I pinged my Twitter friend to see if the
> > same's true for him. I'll try the system upgrade and see if that
> > helps.
> >
> > On 1/17/09, Sean Allen <[hidden email]> wrote:
> >
> > >
> > > On Jan 17, 2009, at 5:33 PM, Giles Bowkett wrote:
> > >
> > >
> > >
> > > > Issue status update for
> > > >
> > > http://smalltalk.gnu.org/project/issue/268
> > >
> > > > Post a follow up:
> > > >
> > > http://smalltalk.gnu.org/project/comments/add/268
> > >
> > > >
> > > > Project:      GNU Smalltalk
> > > > Version:      <none>
> > > > Component:    Build
> > > > Category:     bug reports
> > > > Priority:     normal
> > > > Assigned to:  Unassigned
> > > > Reported by:  swoolerton
> > > > Updated by:   gilesgoatboy
> > > > -Status:       fixed
> > > > +Status:       active
> > > >
> > > > Hi, I changed the status to active because I still see this bug on OS
> X.
> > > > Someone who follows me on Twitter saw it too. We're both on OS X.
> > > > intmath.st fails, rockets to 99+% CPU, and hangs forever. make
> install,
> > > > and running GNU Smalltalk from the command line, succeed anyway.
> > > > MacPorts gives an error message if you try to install GNU Smalltalk
> via
> > > > the port package manager, saying it's unavailable for Intel Macs,
> which
> > > > in 2009 means most Macs, but in fact mine's Intel and it's running
> > > > fine. At least, it handles "Hello world!" and the interactive console.
> > > >
> > > >
> > >
> > > Giles, I installed gst via MacPorts w/o an error message and have run
> both
> > > that version and ones from built from source without issue. I'm using
> > > 10.5.6.
> > >
> > > I just ran intmath.st in less than 10 seconds with both the MacPorts
> > > version
> > > and my built from source version.
> > >
> > > Are you on something earlier than 10.5.6?
> > >
> > >
> > >
> > >
> >
> >
> > --
> > Giles Bowkett
> > http://gilesbowkett.com
> >
>
>


--
Giles Bowkett
http://gilesbowkett.com


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
-Status:       active
+Status:       works for me

No news for a while...




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Giles Bowkett
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   gilesgoatboy
-Status:       works for me
+Status:       active

Well, this is still a problem. I ran into the issue building on OS X
last time, but now I'm building GNU Smalltalk on Ubuntu on my Slicehost
slice and seeing it again. I think last time I was building from
MacPorts, this time I'm building straight from the tarball. Anyway, I
have a feeling this might be a known bug, because I'm having much more
trouble with gst on Ubuntu than I had on OS X, but fwiw, it's very
frustrating.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

toddk42
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   hoohah42
 Status:       active

I have the same issue with 3.1.

Building on Ubuntu 9.0.4, make check hangs on intmath.st .
I cancel with control-C and then end the lt-gst process and everything
proceeds ok.

Further info:

  I am running Ubuntu on a x86 VM under VirtualBox.
  The underlying host hardware is AMD 64 bit Phenom.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
 Status:       active

You don't have libgmp-devel installed, if you compiled your own GNU
Smalltalk (the package name may not be exactly libgmp-devel as it
depends on the distro).  The test would finish in 30-40 minutes.  This
has been lowered in the current development version.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
-Status:       active
+Status:       works for me

Changing status to works for me, the "hang" was fixed and the failure
was never reproduced.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Richard Rasu
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   rasurc
-Status:       works for me
+Status:       active

Just downloaded the 3.1 tarball. Same test as others report here
(intmath.st) hangs with similar issues (100% cpu, ctrl-c not working).
OS X 10.5.8




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk
Reply | Threaded
Open this post in threaded view
|

Re: [bug] intmath.st check fails

Paolo Bonzini-3
In reply to this post by Stephen-71
Issue status update for
http://smalltalk.gnu.org/project/issue/268
Post a follow up:
http://smalltalk.gnu.org/project/comments/add/268

 Project:      GNU Smalltalk
 Version:      <none>
 Component:    Build
 Category:     bug reports
 Priority:     normal
 Assigned to:  Unassigned
 Reported by:  swoolerton
 Updated by:   bonzinip
-Status:       active
+Status:       fixed

Unless you have GMP's header files installed, please do not reopen
issues.  It is fixed in the current trunk, _not_ in the last released
version.




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk