Smalltalk › Squeak › Squeak - Dev

First Environment bug report!

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

3 messages Options

Frank Shearar-3

First Environment bug report!

Try evaluate "Timestamp now" and you'll see an MNU because Environment
doesNotUnderstand: #keysDo:.

It's not a general problem because "Number readFrom: '1'" works as expected.

Oh, that's because I should evaluate TimeStamp, not Timestamp. So this
is something triggered by an UndeclaredVariable, where the Parser's
trying to correct a misspelling. The process hits Class >>
#possibleVariablesFor:continuedFrom: and borks because "self
environment" is insufficiently Dictionary-like.

The fix is easy enough: add #keysDo: and #associationsDo: to
Environment. I'm just writing a test first demonstrating the fix
before I commit anything.

frank

Colin Putney-3

Re: First Environment bug report!

On Fri, Jul 27, 2012 at 5:28 AM, Frank Shearar <[hidden email]> wrote:
> The fix is easy enough: add #keysDo: and #associationsDo: to
> Environment. I'm just writing a test first demonstrating the fix
> before I commit anything.

Great, thanks. I've moved your fix into the trunk so people don't
stumble over the bug in the meantime.

The general strategy I'll pursue for this sort of thing is to create a
new protocol for dealing with Environments, at a higher level of
abstraction than the Dictionary protocol. Come to think of it, we
should deprecate all the dictionary methods on Environment right off
the bat. That'll provide compatibility while we complete the
transition, but make it clear how Environments should be used.

Colin

Frank Shearar-3

Re: First Environment bug report!

On 27 July 2012 18:18, Colin Putney <[hidden email]> wrote:

> On Fri, Jul 27, 2012 at 5:28 AM, Frank Shearar <[hidden email]> wrote:
>> The fix is easy enough: add #keysDo: and #associationsDo: to
>> Environment. I'm just writing a test first demonstrating the fix
>> before I commit anything.
>
> Great, thanks. I've moved your fix into the trunk so people don't
> stumble over the bug in the meantime.
>
> The general strategy I'll pursue for this sort of thing is to create a
> new protocol for dealing with Environments, at a higher level of
> abstraction than the Dictionary protocol. Come to think of it, we
> should deprecate all the dictionary methods on Environment right off
> the bat. That'll provide compatibility while we complete the
> transition, but make it clear how Environments should be used.

That sounds like an excellent plan.

frank

> Colin
>