Old version of XMLSupport is included in GLASS

[jbrichau]

When running the GsUpgrader, I ran into the following load conflict, which indicates GLASS is referencing version 1.0.5 of XMLSupport for GemStone, while version 1.2.2 is stable for use. I have been using it in production for years now.

'Load Conflict between existing ConfigurationOfXMLSupport 1.2.2 from
http://www.squeaksource.com/XMLSupport and ConfigurationOfXMLSupport
1.0.5 from http://seaside.gemtalksystems.com/ss/MetacelloRepository'
Error Category: [User] Number: 1 Arg Count: 1
Arg 1: Load Conflict between existing ConfigurationOfXMLSupport 1.2.2 from http://www.squeaksource.com/XMLSupport and ConfigurationOfXMLSupport 1.0.5 from http://seaside.gemtalksystems.com/ss/MetacelloRepository

[dalehenrich]

Okay ... I looked at the configuration from http://www.squeaksource.com/XMLSupport and according to the config, 1.1.8 is the stable version for GemStone ... I'll change GLASS to use the #stable version and load it from http://www.squeaksource.com/XMLSupport ... I'll also arrange to make sure that XMLSupport and friends are tested ...

[dalehenrich]

Undefined symbols that now show up:

--transcript--'Undefined Symbol:#''OrderedCollectionInspector'''
--transcript--'Undefined Symbol:#''ByteString'''
--transcript--'Undefined Symbol:#''CharacterSetComplement'''
--transcript--'Undefined Symbol:#''WideCharacterSet'''
--transcript--'Undefined Symbol:#''DictionaryInspector'''

sent but not implemented:

--transcript--'Sent but not implemented: #''leadingChar:code:'''

I'll need to clean these up before we're done ...

[dalehenrich]

Name: ConfigurationOfXMLWriter-dkh.22
Author: dkh
Time: 10/11/2014, 18:14:09
UUID: ba27ab24-71db-4ed3-8141-e0e379662800
Ancestors: ConfigurationOfXMLWriter-StephaneDucasse.21

release XMLWriter 1.0.5.1
- bugfix for https://github.com/glassdb/glass/issues/39

and

Name: ConfigurationOfXMLSupport-dkh.66
Author: dkh
Time: 10/11/2014, 18:19:36
UUID: 64c0d144-8e27-46c0-a50b-c05883347aef
Ancestors: ConfigurationOfXMLSupport-dkh.65

release XMLSupport 1.2.2.1 (part two)
- GemStone bugfix for https://github.com/glassdb/glass/issues/39
- depends upone XMLWriter 1.0.5.1
- update Tests groups with test package from XMLWriter

[dalehenrich]

I've updated the configurationOfxml... to 1.2.2 plus XMLWriter and cleaned up the sent but not implemented and undefined globals ... and after loading the XMLS parser code the tests are passing ... then I noticed that there is github xml parser project perhaps I should have referenced the baseline there? ... also the XML tests aren't all passing ...

[jbrichau]

The github xml parser project is the future replacement for XMLSupport yes. But I'm slowly grinding through the failing tests there, so it's not yet for consumption.

I just noticed that 1.2.2 was indeed never tagged stable for gemstone. I just happened to pull in that version for years now in Yesplan and never really noticed. (minor oops :)

[jbrichau]

As I'm trying to fix the errors, I see two things which are fundamentally different in GemStone:

• the XMLStubListObserver uses thisContext in the test case functionality
• the new: aSize method on OrderedCollection in GS behaves differently than Pharo/Squeak and this breaks some tests
  Do you have an idea how to solve the first issue? Is there anything that gives access to the stack context in GS?

[dalehenrich]

I assume you're talking about XMLStubListObserver>>recordNotification? The
following should give you the information that you need:

  | senderFrame senderArgs ms |
  senderFrame := GsProcess _frameContentsAt: 2.
  senderArgs := senderFrame size = 10
    ifTrue: [ #() ]
    ifFalse: [ senderFrame copyFrom: 11 to: senderFrame size ].
  ms := MessageSend
    receiver: (senderFrame at: 10)
    selector: (senderFrame at: 1) selector
    arguments: senderArgs.

thisContext would be GsProcess _frameContentsAt: 1 so #sender is
GsProcess _frameContentsAt: 2...

We are discussing internally what to do about thisContext ... the
frameContents is not the friendliest of interfaces..

Dale

Dale

On Sun, Oct 12, 2014 at 6:30 AM, Johan Brichau notifications@github.com
wrote:

As I'm trying to fix the errors, I see two things which are fundamentally
different in GemStone:

• the XMLStubListObserver uses thisContext in the test case
  functionality
• the new: aSize method on OrderedCollection in GS behaves differently
  than Pharo/Squeak and this breaks some tests Do you have an idea how to
  solve the first issue? Is there anything that gives access to the stack
  context in GS?

—
Reply to this email directly or view it on GitHub
glassdb#39 (comment).

[dalehenrich]

merged to master

[jbrichau]

thx @dalehenrich
I'm going through the remaining failing tests now.

I just want to get that XMLSupport package to the point where we can rely on it. Then it's on to the XMLParser package on github to replace it.

[dalehenrich]

Ah good .... I'm nearing the end of my metacello campaign and I guess I
should get busy on moving glassdb/glass to GsDevKit ... the perhaps I
should focus on Zodiac?

On Fri, Oct 17, 2014 at 11:51 AM, Johan Brichau notifications@github.com
wrote:

thx @dalehenrich https://github.com/dalehenrich
I'm going through the remaining failing tests now.

I just want to get that XMLSupport package to the point where we can rely
on it. Then it's on to the XMLParser package on github to replace it.

—
Reply to this email directly or view it on GitHub
glassdb#39 (comment).