Instance of FFIVariableArgument did not understand #emitArgument:context:

[pavel-krivanek]

The current version has troubles with the TTF initialization.

To reproduce just install the latest mars-gtk and save the image.

FFIVariableArgument(Object)>>doesNotUnderstand: #emitArgument:context:
[ :each | each emitArgument: builder context: sender ] in TFCalloutMethodBuilder>>generateFFICallout:spec: in Block: [ :each | each emitArgument: builder context: send...etc...
OrderedCollection>>do:
TFCalloutMethodBuilder>>generateFFICallout:spec:
[ :builder | self generateFFICallout: builder spec: functionSpec ] in TFCalloutMethodBuilder(FFICalloutMethodBuilder)>>generateMethodFromSpec: in Block: [ :builder | self generateFFICallout: builder spec...etc...
IRBuilder class>>buildIR:
TFCalloutMethodBuilder(FFICalloutMethodBuilder)>>generateMethodFromSpec:
TFCalloutMethodBuilder(FFICalloutMethodBuilder)>>generate
TFCalloutMethodBuilder(FFICalloutMethodBuilder)>>build:
TFCalloutAPI(FFICalloutAPI)>>function:library:
FreeTypeFace(Object)>>ffiCall:library:options:
FreeTypeFace(Object)>>ffiCall:library:
FreeTypeFace(Object)>>ffiCall:
FreeTypeFace(FT2Face)>>ffiOutline:embolden:

[pavel-krivanek]

caused by Pharo commit b839d66, threaddedFFI needs to be synchronized with the UFFI API

[guillep]

Hi Pavel, would you mind checking with the changes in #21?

[pavel-krivanek]

It is not enough. I did a same hotpatch of my image but, at least, I had to define FFICalloutMethodBuilder>>#stringEncodingStrategy too. I did it like

stringEncodingStrategy 
	^ FFILegacyStringEncodingStrategy new

...but I haven't studied how much correct it really is (but it seems to work)

[guillep]

That seems strange. I haven't tried setting TFFI for the whole image (which seems to be what you have...). Gimme some more minutes :)

[guillep]

I've downloaded Mars-gtk in an image with my patch and It seems to be working. My steps:

• load new pharo8
• load TFFI with my patch
• load mars-gtk
• save as Pharo.1.image
• open Pharo.1.image

Did you do something else?

[guillep]

Did you use self requestor as argument of the inCallout: keyword?

[pavel-krivanek]

you are probably right, It seems to work. Let's merge it and I will let rebuild our full images to see if it is really fine. Thanks!

[guillep]

Cool, ping me if you find something else!

[pavel-krivanek]

it looks fixes. Once again, thank you!