[3.9] 7033 available
hilaire at ext.cri74.org
Wed May 31 22:05:53 UTC 2006
Just responding to myself. I tried to send a MC diff to
source.squeakfoundation.org/39a, just got an obvious authorization error
Hilaire Fernandes a écrit :
> I did fixes so the ComplexTest is all green, where should I sent the
> Stéphane Ducasse a écrit :
>>Thanks the network and IO team... please continue this is great work....
>>Even if we nearly lost it because of lost emails :)
>>Time: 21 October 2005, 11:58:31 am
>>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
>>Time: 29 May 2006, 9:29:37 pm
>>Ancestors: Network-KLC.27, Network-tb.28, Network-tb.29
>>Network-tb-29, tb-28, KLC27 with netweork-md.27
>>Time: 24 May 2006, 10:23:58 pm
>>Fixes to address http://bugs.impara.de/view.php?id=2107 FileDirectory
>>class>>retrieveMIMEDocument: class non-existent MIMEType
>>Time: 24 May 2006, 9:10:03 pm
>>Fix for http://bugs.impara.de/view.php?id=2454 - correctly set accent-
>>type in HttpUrl retrieveContents so fetches of .css files succeed
>>Time: 3 April 2006, 11:18 am
>>Ancestors: Network-md.26, Network-gk.18, Network-KLC.21, Network-gk.23
>>This is a merge of:
>>- 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
>>- 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
>>- Did not add isSuperSwikiUrl, it does not belong there. It is a hack
>>from the start.
>>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.
>>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
>>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.
>>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
>>Better class comments in ProtocolClient, Socket and SocketStream.
>>Improved comment in Socket>>closeAndDestroy:'
>>Time: 9 May 2006, 4:54:21 pm
>>- fix loading traits with empty trait composition
More information about the Squeak-dev