Xtreams up to date

Used your Gofer script in Pharo-core 1.2 and Pharo-dev 1.1.,
but after loading I run the tests with many errors and failures...

Is it Squeak only? (which I doubt since the loading script uses
Gofer instead of Installer and Gofer is Pharo while Installer is
Squeak).

Would be nice if we could have a working ConfigurationOfXtream in http://squeaksource.com/MetacelloRepository so it is easy accessible and working packages are versioned together using Metacello.

Thx
T.
--
NEU: FreePhone - kostenlos mobil telefonieren und surfen!
Jetzt informieren: http://www.gmx.net/de/go/freephone

On 12 Jan 2011, at 10:44, Torsten Bergmann wrote:

I see that Martin Kobetic has been doing some checkins lately, maybe you should ask him.
I would be nice if Xtreams works well on Pharo.

Sven

"Torsten Bergmann"<[hidden email]> wrote:
> Used your Gofer script in Pharo-core 1.2 and Pharo-dev 1.1.,
> but after loading I run the tests with many errors and failures...

Hm, that is odd. Maybe I need to try a fresh load and see. I normally get a single failure in one of the socket tests which we haven't tracked down yet. Can you give me a bit more about the kind of errors and failures you see. Does it look like stuff is completely missing ? Maybe I messed up the publishing after the various class moves that I did recently.

> Is it Squeak only? (which I doubt since the loading script uses
> Gofer instead of Installer and Gofer is Pharo while Installer is
> Squeak).

I'm currently using Pharo OneClick image version 1.1.1 directly from the download page. Currently I'm running using the old VM as Cog once crashed on me and I lost a lot of changes. But both VMs seem to run the tests OK.

> Would be nice if we could have a working ConfigurationOfXtream in http://squeaksource.com/MetacelloRepository so it is easy accessible and working packages are versioned together using Metacello.

Yes, I probably need to do that as well. So far I was a bit discouraged mostly by my ignorance. I'm not up to speed on metacello yet, there don't seem to be any tools pre-loaded in the one click image, and I don't know what should one load to get some. I was able to figure out how to use ConfigurationOfFFI but in the end I ended up running the class side load method by hand. The other issue I had was that it seems a metacello configuration requires a bit of maintenance, it seems like you have to manually update the versions when you publish something new. That may be less of a problem now after the updates are done. The Gofer expression seemed a lot easier, and I hoped sufficient for now. Anyway, unless someone beats me to it I'll see what I can do about creating a configuration.

Martin

Den 12.01.2011 14:25, skrev Sven Van Caekenberghe: Many is perhaps an overstatement, when I loaded into a 1.2dev image, all
failures/errors where in relation to XTSocketReadWriting, and many of
those passed if ran independently/restarted in the debugger...

Can't just load the CoreTests and have all tests pass though, due to
the  isAbstract implementation in  XTReadingWritingTest.
Maybe XTReadingWritingTest, XTInfiniteReadingWritingTests and
XTFiniteReadingWritingTests should be moved to the Terminals-Tests package?

Cheers,
Henry

[hidden email] wrote:
> Hm, that is odd. Maybe I need to try a fresh load and see. I normally get a single failure in one of the socket tests which we haven't tracked down yet. Can you give me a bit more about the kind of errors and failures you see. Does it look like stuff is completely missing ? Maybe I messed up the publishing after the various class moves that I did recently.

No, fresh load into one-click 1.1.1 works for me. I guess I need to try different versions.

"Henrik Sperre Johansen"<[hidden email]> wrote:
> Can't just load the CoreTests and have all tests pass though, due to
> the  isAbstract implementation in  XTReadingWritingTest.
> Maybe XTReadingWritingTest, XTInfiniteReadingWritingTests and
> XTFiniteReadingWritingTests should be moved to the Terminals-Tests package?

I could go both ways on this one. They are in CoreTests because the tests themselves test core functionality, but you can't run any of those without the Terminals, so I can see the argument your way too. I wonder though if the move buys us much. There won't be much left in CoreTests if we move the tests that need terminals, it will be basically just the buffer tests. So running the CoreTests without the rest won't tell you much about the state of the port.

Thanks for the additional details on the test failures.

Cheers!

Martin

[hidden email] wrote:
> [hidden email] wrote:
> > Hm, that is odd. Maybe I need to try a fresh load and see. I normally get a single failure in one of the socket tests which we haven't tracked down yet. Can you give me a bit more about the kind of errors and failures you see. Does it look like stuff is completely missing ? Maybe I messed up the publishing after the various class moves that I did recently.
>
> No, fresh load into one-click 1.1.1 works for me. I guess I need to try different versions.

Hm, I'm getting the same result with PharoCore-1.2beta-12294 running on Cog VM. I simply copied the image into my one-click installation, hope that's correct way to do it. Makes me wonder if this is OS dependent problem. I'm running on Fedora 14 (x86_64).

Stuff you probably just didn't bother to write in detail: you would either rename the image and changes or edit the shell script to run the desired image.  I'm pretty sure I have done both to one-click installations, though it was "long" before Cog.  As long as the image has the Cog changes installed (a 1.2 will), it should work.





________________________________________
From: [hidden email] [[hidden email]] On Behalf Of [hidden email] [[hidden email]]
Sent: Wednesday, January 12, 2011 11:30 AM
To: [hidden email]
Cc: Torsten Bergmann; [hidden email]
Subject: Re: [Pharo-project] Xtreams up to date

[hidden email] wrote:
> [hidden email] wrote:
> > Hm, that is odd. Maybe I need to try a fresh load and see. I normally get a single failure in one of the socket tests which we haven't tracked down yet. Can you give me a bit more about the kind of errors and failures you see. Does it look like stuff is completely missing ? Maybe I messed up the publishing after the various class moves that I did recently.
>
> No, fresh load into one-click 1.1.1 works for me. I guess I need to try different versions.

Hm, I'm getting the same result with PharoCore-1.2beta-12294 running on Cog VM. I simply copied the image into my one-click installation, hope that's correct way to do it. Makes me wonder if this is OS dependent problem. I'm running on Fedora 14 (x86_64).