From: St�phane Ducasse <stephane.ducasse@inria.fr>

Date: November 6, 2010 10:01:35 PM GMT+01:00

To: Lukas Renggli <renggli@me.com>

Subject: Strange behavior and need help

Hi lukas

if you have a couple of minutes could you have a look at the following?

I wanted to port coral to the latest AST, petit parseur....

So I was expected to have my test green since I hardcoded version of AST, petit*

When I tried to load coral in 1.1-rc4 I got some red tests.

I compared the tests are the same and Coral code too.

I did several experiments:

In 1,1rc4 I loaded

Gofer new

squeaksource: 'Coral';

package: 'ConfigurationOfCoral';

load.

(ConfigurationOfCoral project version: '0.1') load

*do not load this one**

I have images where the tests are all green. But they predate 1.1rc4 :(

Of course I load really old versions of Petit* and AST but I do not understand why I cannot get them working.

May be the compiler changed.

in PharoCore1.1ALPHA 11131 or in an image 1.1-11203

I have

Gofer new squeaksource: 'rb'; version: 'AST-Core-lr.35'; load.

Gofer new renggli: 'petit'; version: 'PetitParser-lr.122' ; version: 'PetitSmalltalk-lr.13'; load.

Gofer new squeaksource: 'Coral'; package: 'Coral'; load.

and the tests are green

when I load that in a 1.1 rc I get this test

testWhiteSpaceAfterClassName failing

So my problem is that two guys (one admin used to program and write books on rebol are starting to invest time to build coral

and I do not have time for fixing that now) and I'm now in a situation where I cannot reproduce an image. Shit I should have migrated verything to 1.1 before.

So could you have a look because with the latest version of AST, PetitParser, PetitSmalltalk

it may take you some minutes to fix that.

I produced a new configuration with your latest version.

Gofer new

squeaksource: 'Coral';

package: 'ConfigurationOfCoral';

load.

(ConfigurationOfCoral project version: '0.2') load

I started to work but I could not understand why space got an influence.

testMethodBinary3

"self debug: #testMethodBinary3"

self

parse: ' Point >> + arg

[^ self + arg ] '

rule: #method

is red and

testMethodBinary2

self parse: ' Point >> + arg [^ self + arg ]'

rule: #method

is green

Thanks in advance.

Stef

Si vous r�pondez � ce message, en laissant l'en-t�te Objet: intact,
Mailman supprimera le message en attente. Proc�dez ainsi s'il s'agit
d'un spam. Si vous r�pondez � ce message en y ajoutant un en-t�te
Approuv�: avec le mot de passe de la liste comme valeur, le message
sera autoris� � aller vers la liste. L'en-t�te Approuv�: peut
�galement appara�tre � la premi�re ligne du corps du message de
r�ponse