[squeak-dev] Xtreams doesn't load into Squeak trunk

Frank Shearar frank.shearar at gmail.com
Sun Apr 8 10:44:02 UTC 2012


On 8 April 2012 00:22, Colin Putney <colin at wiresong.com> wrote:
>
> On 2012-04-07, at 2:49 PM, Frank Shearar wrote:
>
>> I'll take a further look into the issue, but in my clean #11995 I just
>> tried to load Xtreams:
>>
>> Installer ss
>>       project: 'MetacelloRepository';
>>       install: 'ConfigurationOfXtreams'.
>> (Smalltalk at: #ConfigurationOfXtreams) project bleedingEdge load
>>
>> and I get a debugger saying there's an MNU: an ExternalFunction
>> class(Object) doesn't understand #callingConventionModifierFor: in
>> Parser>>externalFunctionDeclaration.
>>
>> That method's been around a while - eem touched it last back in 2009,
>> which is well before I first played with Xtreams [1] - so it's not
>> like strange new behaviour landed in the image.
>>
>> I can see that the ConfigurationOf pulls in FFI - FFI-Kernel-tbn.20,
>> FFI-Pools-eem.2 are loaded at this point - and it's during the loading
>> of FFI-Tests-jcg.4 that I hit a problem.
>>
>> I wonder if anyone else can replicate the issue?
>
> IIRC, ConfigurationOfXtreams specifies an old version of FFI - one that doesn't work in recent versions of Squeak. You can either load the latest FFI instead, or use Xtreams without Xtreams-Xtras, which is the part that depends on FFI.

Thanks! Yes, that's exactly it. FFI-Kernel-eem.24 loads the method I
need. It looks like my 4.3-gamma-upgraded-to-11925 image thinks it's
still a Squeak 4.2 (judging by SystemVersion), and so tries to load an
old spec for XTreams.

If I confirm that (by installing into a freshly downloaded 4.4-alpha),
I'll submit a corrected ConfigurationOfXTreams that can load without
pulling packages down the wire that are already in the local
package-cache.

I'm still running the tests, but it looks like "(Smalltalk at:
#ConfigurationOfXtreams) project development load" is the correct
incantation. Ah, they finished. 850 run, 848 passed!

XTFileReadingWritingTest>>#testWriteTransformingOverLimitingPastEnd
fails in an intermittent manner, so I can't tell why it failed the
first time. (It'd be really great if we had some logging mechanism
connected to the TestRunner!)

XTSocketReadingWritingTest>>#testReadWriteLargeAmount fails because
self assert: serverSuccess fails.

OK, so I'll see to updating ConfigurationOfXtreams.

frank

> Colin


More information about the Squeak-dev mailing list