search

Metacello / metacello

Metacello is a package management system for Smalltalk
MIT License
87 stars 43 forks source link

add pharo-6.1 to lineup #448

Closed dalehenrich closed 7 years ago

dalehenrich commented 7 years ago

Looks like same 13 tests fail in Pharo6.0

dalehenrich commented 7 years ago

Also looks like @estebanlm is addressing the test failures with his work in PR #452

estebanlm commented 7 years ago

@dalehenrich I have no idea why Pharo 5.0 tests are failing, and I cannot reproduce it in my machine :( any idea?

krono commented 7 years ago

Well, 

Smalltalk 
        at: #SessionManager
        ifPresent: [ :c | c default registerNetworkClassNamed: self name ]
        ifAbsent:[ Smalltalk addToStartUpList: self ].

broke all builds...

dalehenrich commented 7 years ago

@estebanlm I agree with @krono it certainly looks like #at:ifPresent:ifAbsent: is the culprit ... perhaps in your image, MCGitBasedNetworkRepository is already initialized, since that is where the failure appears to be happening on travis?

krono commented 7 years ago

Or Pharo has SmalltalkImage>> #at:ifPresent:ifAbsent:

krono commented 7 years ago

Also, it would work with Smalltalk globals #at:ifPresent:ifAbsent: in Squeak..

estebanlm commented 7 years ago

hmmm... ok, I will submit a fix tomorrow.

estebanlm commented 7 years ago

and now I screwed the Pharo platform code. this thing of monticello branches is annoying.

krono commented 7 years ago

What about 

Smalltalk globals
        at: #SessionManager
        ifPresent: [ :c | c default registerNetworkClassNamed: self name ]
        ifAbsent:[ Smalltalk addToStartUpList: self ].
estebanlm commented 7 years ago

meh… maybe… I took #classNamed: because that’s what we have for FFI-Kernel. but in the merge I lost the baseline… grrr… anyway, retrying :)

On 28 Sep 2017, at 14:44, Tobias Pape notifications@github.com wrote:

What about

Smalltalk globals at: #SessionManager ifPresent: [ :c | c default registerNetworkClassNamed: self name ] ifAbsent:[ Smalltalk addToStartUpList: self ]. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Metacello/metacello/pull/448#issuecomment-332824635, or mute the thread https://github.com/notifications/unsubscribe-auth/AAfWXjMCSZ1SoG7G_rfO8i2w08KEiIZVks5sm5SmgaJpZM4PNAMB.

estebanlm commented 7 years ago

still failing, now for other reasons :( I will check what I broke tomorrow.

estebanlm commented 7 years ago

So. Now, I don't know why Gemstone is failing :) (also, the cache problem is becoming... annoying)

estebanlm commented 7 years ago

yes I saw that... but I did not make that change, that's why is weird.

dalehenrich commented 7 years ago

I've commented on your checking ... #includesAll: is not portabel ... #includesAllOf: is portable ... or at least used to be ...

estebanlm commented 7 years ago

mmm... #includesAllOf: was victim of autodeprecation tool and I didn't see it :( yet another issue to fix :'(

dalehenrich commented 7 years ago

With regards to the "cache problem", I assume you are referring to this? I consider clearing the cache a workaround to a "pharo problem"

Have you looked into who is doing the mv? This seems to be a bug in the smalltalkCI pharo download code somewhere ...

dalehenrich commented 7 years ago

Yeah, writing portable code is real pain when one or more platforms decide to deprecate formerly portable methods ... it's why ANSI Smalltalk was created ... it =should be possible for a Smalltalk programmer to write portable code with the expectation that the lifetime of that portable code is greater then 1 year ... Perhaps a new effort at redefining what should be considered as portable should be started?

estebanlm commented 7 years ago

yeah... annoying. But do not blame me! While I can agree with changes that improves the platform, I dislike the name changes "just because"... (in this case it was because there were two equivalent methods and who chose which one to keep didn't know about compatibility :( )

dalehenrich commented 7 years ago

I'm not blaming, but I am suggesting that a stake be put into the ground for a set of protocol that can be depended upon to not change between platforms and for the foreseeable future :)

estebanlm commented 7 years ago

ok, cache problem aside, seems that this time I nailed it (finally!)

dalehenrich commented 7 years ago

I've cleared the caches and restarted the pharo6.1 build for the PR and the build on your the source branch ... they should both be green before merging a pull request ...

dalehenrich commented 7 years ago

@estebanlm this puppy dog is ready to go ... you want to merge the pull request and exercise your owner super powers?

estebanlm commented 7 years ago

ok :)