New subject: which fix is the best for Complex?

29 May 2006


      Thanks the network and IO team... please continue this is great work....
Even if we nearly lost it because of lost emails :)
Stef
Name: System-gk.45
Author: gk
Time: 21 October 2005, 11:58:31 am
UUID: f8210799-b1de-5f46-b675-8c7b83550503
Ancestors: System-gk.44
Two good refactorings based on work from Bernard Pieber for Mantis  
#861 in the Network package. These depend on changes in the Network  
package:
- Project class>>fromUrl:
- ProjectLoading class>>bestAccessToFileName:andDirectory:
This also removes the only sender of Project  
class>>serverFileFromURL: and we could nuke it, but we leave that  
decision for the System stewards. :)"
"Name: Network-sd.30
Author: sd
Time: 29 May 2006, 9:29:37 pm
UUID: 6da57032-4ee0-412e-830a-e6169251929e
Ancestors: Network-KLC.27, Network-tb.28, Network-tb.29
merge
Network-tb-29, tb-28, KLC27 with netweork-md.27
Name: Network-tb.29
Author: tb
Time: 24 May 2006, 10:23:58 pm
UUID: dfdd5a1c-4494-41da-afff-925a1597b1c3
Ancestors: Network-md.27
Fixes to address http://bugs.impara.de/view.php?id=2107   
FileDirectory class>>retrieveMIMEDocument: class non-existent  
MIMEType class>forExtension:
Name: Network-tb.28
Author: tb
Time: 24 May 2006, 9:10:03 pm
UUID: 75f7e92a-6087-4223-a373-9d6516cd35e1
Ancestors: Network-md.27
Fix for http://bugs.impara.de/view.php?id=2454 - correctly set accent- 
type in HttpUrl retrieveContents so fetches of .css files succeed
Name: Network-KLC.27
Author: KLC
Time: 3 April 2006, 11:18 am
UUID: b2510551-ffee-4449-94c5-48d8c31bef77
Ancestors: Network-md.26, Network-gk.18, Network-KLC.21, Network-gk.23
This is a merge of:
Network-gk.18.mcz
Network-KLC.21.mcz
Network-gk.23.mcz
Combined notes:
Network-gk.18:
Changes:
- Adds #fileName to FileUrl and HierarchicalUrl (with HttpUrl as  
subclass), ok - I think that is fair even if the last part of the  
path not necessarily denotes a file.
- Adds Url class>>directoryUrl to FileUrl and HierarchicalUrl (with  
HttpUrl as subclass) which simply returns a new URL with the last  
element in the path removed.
- Adds #absoluteFromFileNameOrUrlString: which then is used from two  
different places (refactoring).
- A fix so that copying a URL does not share the path collection with  
the original. Not sure exactly if this has caused an actual bug yet,  
but I agree it is a good idea to not share.
- Fix bug in FileUrl>>pathForDirectory which used an explicit  
pathname delimiter in one place. Duh.
- Removed class MswUrl, sorry but this should not be in standard  
Network package IMHO. There is no use of it in the current image.
Non behavior changes:
- Refactoring adding Url class>>urlClassForScheme:.
- Improved naming in ServerDirectory class>>serverForURL:, no actual  
code change.
- Refines and adds lots of comments here and there.
Differences from proposed changes from Bernard Pieber in Mantis #861:
- Did not include the proposed HierarchicalPath>>pathForDirectory  
because it doesn t return a path (like the same method in FileUrl  
does), instead it returns a URL turned into a String. I changed the  
few senders that Bernard Pieber (Mantis #861) introduced to  
directoryUrl asString instead.
- Did not add isSuperSwikiUrl, it does not belong there. It is a hack  
from the start.
Network-KLC.19:
This is an adaptation of the updated provided by Patrick Mauritz at  
Mantis ID #0411:
Change Set: Enh-ConnectionQueue
Date: 16 October 2004
Author: Patrick Mauritz
Allow classes to subscribe to the ConnectionQueue via addDependent:  
to get notifications when new connections are opened.
Receivers need to implement update:aSocket for this to work.
Thanks to Ken Causey for guidance on IRC.
I adapted Patrick s changes to ConnectionQueue>>listenLoop the result  
being that a check for a ConnectedTimedOut exception is added and a  
call to self changed on successful connection. Otherwise this should  
be the same as previously in the image.
Network-KLC.20:
As reported in Mantis ID #2106 this method included a call to the now  
non-existent Notification class>deprecated method. I ve replaced that  
call. With what I think is the appropriate mechanism.
The next question is at what point do deprecated methods get removed  
entirely?
Network-KLC.21:
This removes 3 methods that had no implementation other than calls to  
self halt. None of the 3 methods were being used and it didn t appear  
that they would likely ever be used.
Network-gk.23:
Fix for Mantis #2119, the simplest one described.
Fix for Mantis #1585, moved 3 test methods from OldSocket class to  
Socket and updated the code so that it works.
Doing the above I also fixed an issue in
Socket>>waitForDisconnectionFor:, it was wrongly waiting on semaphore  
instead of self readSemaphore. I also removed an unnecessary call to  
dataAvailable, unneeded temp var #extraBytes etc. The problem was  
that loopbackTest randomly hung in this method.
Optimized Socket>>discardReceivedData, unneeded call to dataAvailable  
removed.
Better class comments in ProtocolClient, Socket and SocketStream.
Improved comment in Socket>>closeAndDestroy:'
Name: Monticello-al.297
Author: al
Time: 9 May 2006, 4:54:21 pm
UUID: bc893983-a6cd-430c-bf5d-a4d60fdb04b3
Ancestors: Monticello-al.295
- fix loading traits with empty trait composition

[3.9] 7033 available