[UPDATES] 7 for 3.6beta, including "full" prompt
Doug Way
dway at riskmetrics.com
Mon Aug 25 05:31:27 UTC 2003
These are the latest approved fixes (which are not for 3.7alpha), plus
the latest SARInstaller w/fixes, and the much-discussed "upgrade to
full" prompt.
- Doug
--------------------------
5396BrowserRedrawFix-nk -- Ned Konz -- 6 August 2003
Fixes the redrawing bug in the SystemBrowser (redraw didn't remove the
focus rectangles).
But only invalidates the focus rectangles, so we don't have to do a
full redraw.
Since we're always grabbing keyboard focus on mouseEnter, removed
conditional in mouseLeave. This will prevent a PLM from responding to
keyboard events once the mouse has left its bounds."
5397PrefRemoval-ar -- Andreas Raab -- 25 July 2003
Rename the method #expungePreferenceNamed: to #removePreference: in
order to be consistent with #addPreference:"
5398socketacceptfix-svp -- Stephen Pair -- 30 July 2003
md: made a real changeset...
Per Michael's request...I'm re-submitting this as a fix. It should be
put into the update stream and the network rewrite packages.
- Stephen
Stephen Pair wrote:
> It appears that Socket>>waitForAcceptFor: and friends are not working
> in the latest 3.6 image. Attached is a change set that fixes it and
> also adds a couple of related convenience methods.
>
> Also, in 3.6 there appear to be a number of changes relating to method
> contexts and such...I keep encountering 'Block cannot return' errors
> during process unwinding (on termination)...these errors are happening
> in code that worked fine in previous releases (specifically it's
> happening when stopping a Service in KomServices). Has anyone
> experienced this?
>
> - Stephen"
5399SocketFixRev3-mu -- Masashi Umezawa -- 17 August 2003
After installing Stephen Pair's socket accept fix, I still found minor
bugs. So I fixed them.
-Socket>>receiveAvailableDataInto:startingAt: ansers self.
Now it answers the size of received data correctly.
-Socket>>connectTo:port:waitForConnectionFor:
-Socket>>waitForConnectionFor:
Both try to report bad host address, but it fails. (because remoteHost
is not known if it fails to connect).
Now it reports bad host address and port in timeout.
(added #waitForConnectionFor: from masashi's aggregated changeset -dew)
5400RevertExcDesc2-ab -- Avi Bryant -- 22 August 2003
v.2
Now uses Class>>selector as the message text for MessageNotUnderstood,
after discussion with Doug and Julian.
v.1
One thing that (perhaps irrationally) drives me nuts about the
ContextEnh changes is that, for no apparent reason,
MessageNotUnderstood exceptions now yield 'SmallInteger quux?' as their
description text rather than 'MessageNotUnderstood: quux'. This
changeset simply reverts two methods (Exception>>description and
MessageNotUnderstood>>messageText) back to their original versions.
5401SARInstallerFor36-18 -- Ned Konz -- 22 August 2003
22 August (v18)
* removed prompt for loading DVS
* removed linefeeds
5402PromptUpgradeToFull-dew -- Doug Way -- 25 August 2003
Add a prompt allowing people to upgrade to a Full image if desired."
"First make sure SM is updated"
SMSqueakMap default loadUpdates.
(self confirm:
'With this update, you have the opportunity to ''upgrade'' your image
to the ''Full'' multimedia release for 3.6, if you select Yes. This
will
install these previously removed packages into your image via SqueakMap:
BaseImage Tests, Celeste, Games, VMMaker, Benchmarks, Scamper, and
Balloon3D.
-
You can then continue to follow the update stream with your Full image.
However, the update stream will not update these separate packages; you
will need to update them yourself from SqueakMap if they become out of
date.
(Which you can do by opening a Package Loader and using the
''display installed packages'' filter.)
-
Or, if you select No, you will stay with the slimmed-down ''Basic''
developer
image, and you can follow the update stream that way. This is the
default option.
-
Would you like to load these extra packages to move to a ''Full''
image?')
ifTrue: [SMSqueakMap default installPackageWithId:
'ebce5a15-1916-4bec-a77c-715f6c2ab6e5'].
More information about the Squeak-dev
mailing list
|