I just posted a small fix on the list that we need to include in 3.4 right away - otherwise you can't change the class definition on the class side! Which seems bad. ;-)
The bug was introduced by Torge's 5137 fix.
regards, Göran
"Andreas Raab" andreas.raab@gmx.de wrote:
The bug was introduced by Torge's 5137 fix.
What was the fix?! If I revert that change then everything seems to be all right. It doesn't have any effect I can see...
It appeared to be a check so that one can't enter methods when the method category "all" is selected. But I never verified that Torge's check worked... I just assumed it did.
Aha, ok - the idea is that it gives you a proper "inform" instad of "Nothing more expected ->" if you start entering a method without first selecting a method category. If I understood it correctly.
On the other hand it does feel really ugly to test for certain keywords in the message (that is also why the fix turned sour) - my fix is equally ugly in that respect. Actually - I vote for dropping this altogether, it's simply bad code - mine included, but I claim innocense since I got dragged into it by Torge! :-).
regards, Göran
On Mon, 16 Dec 2002 goran.hultgren@bluefish.se wrote:
"Andreas Raab" andreas.raab@gmx.de wrote:
The bug was introduced by Torge's 5137 fix.
What was the fix?! If I revert that change then everything seems to be all right. It doesn't have any effect I can see...
It appeared to be a check so that one can't enter methods when the method category "all" is selected. But I never verified that Torge's check worked... I just assumed it did.
Aha, ok - the idea is that it gives you a proper "inform" instad of "Nothing more expected ->" if you start entering a method without first selecting a method category. If I understood it correctly.
Didn't this used to (way back yonder) create a new protocol (or add to an existing protocol) called "as yet unclassified" for methods entered while "--all--" is selected? What happened to that? (Or did I simply dream that this was the case, in some kind of "perverse hacker nightmare"?)
Ian
Ian Piumarta ian.piumarta@inria.fr wrote:
On Mon, 16 Dec 2002 goran.hultgren@bluefish.se wrote:
"Andreas Raab" andreas.raab@gmx.de wrote:
The bug was introduced by Torge's 5137 fix.
What was the fix?! If I revert that change then everything seems to be all right. It doesn't have any effect I can see...
It appeared to be a check so that one can't enter methods when the method category "all" is selected. But I never verified that Torge's check worked... I just assumed it did.
Aha, ok - the idea is that it gives you a proper "inform" instad of "Nothing more expected ->" if you start entering a method without first selecting a method category. If I understood it correctly.
Didn't this used to (way back yonder) create a new protocol (or add to an existing protocol) called "as yet unclassified" for methods entered while "--all--" is selected? What happened to that? (Or did I simply dream that this was the case, in some kind of "perverse hacker nightmare"?)
Ian
It depends a bit on how you interact... :-) If you select a class, select "--all--" and then enter a method (instead of the class template) then you get the "Nothing more expected ->" thing (and we all know why). This was what Torge wanted to make friendlier with an inform. Noble intention but I don't like the code - not even my fixed version that works...
But if you select a class, select "--all--", select a method, deselect that method (giving you a method template) and enter a new method then it will - as you noted - end up in "as yet unclassified".
In short - the whole Browser-caboodle feels like a big mess sometimes and it would be great to just rewrite the whole darn thing. That's another story of course.
If this "newbie problem" really needs to be "patched" somehow then we could instead throw up some nice inform: when we trap that the Compiler failed to compile the expression. That would be much cleaner IMHO.
regards, Göran
Ian Piumarta wrote:
On Mon, 16 Dec 2002 goran.hultgren@bluefish.se wrote:
Aha, ok - the idea is that it gives you a proper "inform" instad of "Nothing more expected ->" if you start entering a method without first selecting a method category. If I understood it correctly.
Didn't this used to (way back yonder) create a new protocol (or add to an existing protocol) called "as yet unclassified" for methods entered while "--all--" is selected? What happened to that? (Or did I simply dream that this was the case, in some kind of "perverse hacker nightmare"?)
This was changed awhile ago so that you could see the class template and the class' list of methods at the same time, when selecting a class in the browser.
See http://groups.yahoo.com/group/squeak/message/52761 for more background.
I could go either way with either using Goran's fix, or just backing out Torge's enhancement. Goran said he'd vote to back out the enhancement, so maybe we should just do that.
To me, the situation before Torge's enhancement wasn't all that bad... if I see the class template in the browser text pane, I know that I won't be able to accept a method there. Maybe post-3.4, we could make things more newbie-friendly by adding "define class" and "add method" menu items in the appropriate browser menus, which would somehow steer people in the right direction (while still preserving the existing ways of defining classes & adding methods, of course).
- Doug
Methinks that perhaps the real problem is the automatic selection of '--all--' in the browser. If this weren't there, the mechanism would be rather obvious - you select the class you edit the class. You select a method category you edit a method. This is already true *except* for the '--all--' category which attempts to edit the class definition when it is selected.
Cheers, - Andreas
-----Original Message----- From: squeakfoundation-admin@lists.squeakfoundation.org [mailto:squeakfoundation-admin@lists.squeakfoundation.org] On Behalf Of Doug Way Sent: Monday, December 16, 2002 8:54 PM To: squeakfoundation@lists.squeakfoundation.org Subject: Re: [Squeakfoundation]defineClass fix!
Ian Piumarta wrote:
On Mon, 16 Dec 2002 goran.hultgren@bluefish.se wrote:
Aha, ok - the idea is that it gives you a proper "inform"
instad of
"Nothing more expected ->" if you start entering a method
without first
selecting a method category. If I understood it correctly.
Didn't this used to (way back yonder) create a new protocol
(or add to an
existing protocol) called "as yet unclassified" for methods
entered while
"--all--" is selected? What happened to that? (Or did I
simply dream
that this was the case, in some kind of "perverse hacker
nightmare"?)
This was changed awhile ago so that you could see the class template and the class' list of methods at the same time, when selecting a class in the browser.
See http://groups.yahoo.com/group/squeak/message/52761 for more background.
I could go either way with either using Goran's fix, or just backing out Torge's enhancement. Goran said he'd vote to back out the enhancement, so maybe we should just do that.
To me, the situation before Torge's enhancement wasn't all that bad... if I see the class template in the browser text pane, I know that I won't be able to accept a method there. Maybe post-3.4, we could make things more newbie-friendly by adding "define class" and "add method" menu items in the appropriate browser menus, which would somehow steer people in the right direction (while still preserving the existing ways of defining classes & adding methods, of course).
- Doug
Squeakfoundation mailing list Squeakfoundation@lists.squeakfoundation.org http://lists.squeakfoundation.org/listinfo/squeakfoundation
I replied to this over on squeak-dev. When a discussion here on the SqF list moves away from Guides/community/release-related topics to general Squeak technical discussion (or other topics), it's probably good to move it over to squeak-dev.
- Doug
Andreas Raab wrote:
Methinks that perhaps the real problem is the automatic selection of '--all--' in the browser. If this weren't there, the mechanism would be rather obvious - you select the class you edit the class. You select a method category you edit a method. This is already true *except* for the '--all--' category which attempts to edit the class definition when it is selected.
Cheers,
- Andreas
-----Original Message----- From: squeakfoundation-admin@lists.squeakfoundation.org [mailto:squeakfoundation-admin@lists.squeakfoundation.org] On Behalf Of Doug Way Sent: Monday, December 16, 2002 8:54 PM To: squeakfoundation@lists.squeakfoundation.org Subject: Re: [Squeakfoundation]defineClass fix!
Ian Piumarta wrote:
On Mon, 16 Dec 2002 goran.hultgren@bluefish.se wrote:
Aha, ok - the idea is that it gives you a proper "inform"
instad of
"Nothing more expected ->" if you start entering a method
without first
selecting a method category. If I understood it correctly.
Didn't this used to (way back yonder) create a new protocol
(or add to an
existing protocol) called "as yet unclassified" for methods
entered while
"--all--" is selected? What happened to that? (Or did I
simply dream
that this was the case, in some kind of "perverse hacker
nightmare"?)
This was changed awhile ago so that you could see the class template and the class' list of methods at the same time, when selecting a class in the browser.
See http://groups.yahoo.com/group/squeak/message/52761 for more background.
I could go either way with either using Goran's fix, or just backing out Torge's enhancement. Goran said he'd vote to back out the enhancement, so maybe we should just do that.
To me, the situation before Torge's enhancement wasn't all that bad... if I see the class template in the browser text pane, I know that I won't be able to accept a method there. Maybe post-3.4, we could make things more newbie-friendly by adding "define class" and "add method" menu items in the appropriate browser menus, which would somehow steer people in the right direction (while still preserving the existing ways of defining classes & adding methods, of course).
- Doug
Squeakfoundation mailing list Squeakfoundation@lists.squeakfoundation.org http://lists.squeakfoundation.org/listinfo/squeakfoundation
Squeakfoundation mailing list Squeakfoundation@lists.squeakfoundation.org http://lists.squeakfoundation.org/listinfo/squeakfoundation
squeakfoundation@lists.squeakfoundation.org
-
Andreas Raab -
Doug Way -
goran.hultgren＠bluefish.se -
Ian Piumarta