Issue 6097 in pharo: Bad source and changes path for one click on Windows (Jenkins?)

Status: Accepted
Owner: [hidden email]
Labels: Type-Bug Milestone-2.0

New issue 6097 by [hidden email]: Bad source and changes path for one  
click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

When one downloads the one click from Jenkins (Pharo2.0a Latest update:  
#20144)

https://ci.lille.inria.fr/pharo/view/Pharo%202.0/job/Pharo-2.0/lastSuccessfulBuild/artifact/Pharo-2.0-one-click.zip

extract it on Windows under C:\test20 and uses
Pharo-2.0-one-click.bat the source file is not found.

It tries to search the file under:

/builds/jenkins/workspace/Pharo-2.0/Pharo-2.0/c:\test20\Pharo-2.0-one-click.app\/PharoV10.sources

Also the monticello local repository points to a "bad path":

C:\test20\Pharo-2.0-one-click.app\Contents\Resources\C:\test20\Pharo-2.0-one-click.app\Contents\Resources

Looks like file system issues on Win32 or it is coming from
the interaction with Jenkins.


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #1 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

Under OSX 10.7.4 I don't get any errors.. so this must be a Windows only  
issue.
Does the same happen under windows with the latest VM and Image from  
jenkins?


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #2 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

As I wrote I used the Jenkins one (update 20144 is from today).

I assume it is a windows only issue.

Similar to the windows only #6096  
(http://code.google.com/p/pharo/issues/detail?id=6096)

Dont know how well FS was tested on Windows...

Still wonder that you dont have any machine with M$ software around...


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #3 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

usually we test all that stuff with our windows slave on jenkins (if  
jenkins is up and such :/)

I just wanted to know if its a one-click win problem or if it happens as  
well under the the separate image.. (for instance I never ever use onelick  
images :P)


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #4 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

The problem is #fullName: when used with a full path already.

Pharo 1.4

    FileStream fullName: 'c:\foo\test.txt' gives 'c:\foo\test.txt'

Pharo 2.0:

    FileStream fullName: 'c:\foo\test.txt' gives
      'C:\pharo\versions\2.0\Pharo-2.0-one-click.app\Contents\Resources\c:\foo\test.txt'


So the old API is broken and constructs invalid paths.

This breaks lots of other functionality of file streams like:

   FileStream fileNamed: 'x.txt'

returns a concrete filestream (MutibyteFileStream instance) in Pharo 1.4

returns nil in Pharo 2.0 due to the internal problem that
by calling #fullName twice the path gets doubled/gets invalid

   'C:\pharo\versions\2.0\Pharo-2.0-one-    
click.app\Contents\Resources\C:\pharo\versions\2.0\Pharo-2.0-one-click.app\Contents\Resources\test.txt'

If you dive deeper you will notice that

FileSystem disk resolve: 'c:\test.txt'

constructs paths like:

Path * 'C:' / 'pharo' / 'versions' / '2.0' / 'Pharo-2.0-one-click.app'  
/ 'Contents' / 'Resources' / 'c:' / 'test.txt'

since it mixes it with the working directory





_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #5 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

just to confirm: it is not only broken in one click. Its also in the  
regular image cleanly downloaded from CI:

   
https://ci.lille.inria.fr/pharo/view/Pharo%202.0/job/Pharo-2.0/lastSuccessfulBuild/artifact/Pharo-2.0.zip

So it is a general problem.


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #6 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

In an older Pharo 2.0 (Pharo2.0a Latest update: #20119) if you do

     FileSystem disk resolve: 'c:\test.txt'

it is correctly resolved into  Path / 'c:' / 'test.txt'


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #7 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

tracking further

   (WindowsStore current  pathFromString: 'c:\test.txt') class

returns an instance of RelativePath. In the old 2.0 image it returned
correctly an AbsolutePath


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Labels: Importance-High

Comment #8 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

OK, I've found it:

(AbsolutePath from: 'c:\test.txt') class

returns RelativePath on Windows instead of AbsolutePath. This is clearly
a bug and was introduced with "FileSystem-Core-StephaneDucasse.14.mcz"
providing the fixes from issue 5967 (use Filesystem for everything).

The problem is the method Path>>from:delimiter where an AbsolutePath
is only used when the path string starts with "/".

This is true for Unix but NOT FOR WINDOWS path names.

Here are two Sunit tests that should cover the problem,
the second one is red and should be made green:

testUnixAbsolutePathName
   self assert: (AbsolutePath from: '/test') isAbsolute

testWindowsAbsolutePathName
   self assert: (AbsolutePath from: 'c:\test') isAbsolute

Please fix it on a non-windows platform since the image
is unusable on Win32 systems due to this bug.


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #9 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

Also note that there is a bug in the method WindowsStore>>pathFromString:
again with the Absolute/RelativePath distinction since this method
assumes that the given path string has at least a size of two characters.

If you have a relative directory that is called "a" (just one character)  
this
would lead to a Subscript out of bounds error when sending #second.
Try yourself:

   WindowsStore current pathFromString: 'a'


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Status: FixToInclude

Comment #10 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

in  
SLICE-Issue-6097-Bad-source-and-changes-path-for-one-click-on-Windows-Jenkins-EstebanLorenzano.1


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Status: Integrated

Comment #11 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

in 20148


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Status: Workneeded

Comment #12 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

reopen, is having some issues with changes file


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Status: FixToInclude

Comment #13 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

in  
SLICE-Issue-6097-Bad-source-and-changes-path-for-one-click-on-Windows-Jenkins-EstebanLorenzano.4
Author: EstebanLorenzano


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Updates:
        Status: Integrated

Comment #14 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

in 20149


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker

Comment #15 on issue 6097 by [hidden email]: Bad source and changes  
path for one click on Windows (Jenkins?)
http://code.google.com/p/pharo/issues/detail?id=6097

For the other two problems described I created

  http://code.google.com/p/pharo/issues/detail?id=6139 and
  http://code.google.com/p/pharo/issues/detail?id=6138


_______________________________________________
Pharo-bugtracker mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-bugtracker