1. SVN Revision tests failing >= 1910
- Posted by SDPringle Apr 11, 2009
- 902 views
I am seeing in a large proportion of the unit tests the interpreter issues an error that Windows handles by showing me a dialog that I must click on. I have tried this on svn 1910, 1913, and now 1908, 1897, 1851.
I am asked whether to send an error report to Microsoft or not. When running eutest.ex, I will get scores of these blocking dialog boxes unless I user break out of the eutest with CNTRL+C.
Before 1910, only t_command_line_quote.e was failing this way. After 1910, more than just that test.
Shawn
2. Re: SVN Revision tests failing >= 1910
- Posted by mattlewis (admin) Apr 11, 2009
- 875 views
- Last edited Apr 12, 2009
I am seeing in a large proportion of the unit tests the interpreter issues an error that Windows handles by showing me a dialog that I must click on. I have tried this on svn 1910, 1913, and now 1908, 1897, 1851.
I am asked whether to send an error report to Microsoft or not. When running eutest.ex, I will get scores of these blocking dialog boxes unless I user break out of the eutest with CNTRL+C.
Before 1910, only t_command_line_quote.e was failing this way. After 1910, more than just that test.
Yes, it's all the same problem. I'm working out some issues with the new code blocks, which manage reference counting differently than euphoria used to.
Matt
3. Re: SVN Revision tests failing >= 1910
- Posted by Critic Apr 12, 2009
- 933 views
I'm working out some issues with the new code blocks, which manage reference counting differently than euphoria used to.
Watch out that compiler generated temporaries don't increment the reference count. Otherwise you get the poor O(n) vs. O(1) behaviour which I meant weeks ago. (Sorry, I have been busy, so I never read your answer.)
4. Re: SVN Revision tests failing >= 1910
- Posted by mattlewis (admin) Apr 12, 2009
- 923 views
- Last edited Apr 13, 2009
I'm working out some issues with the new code blocks, which manage reference counting differently than euphoria used to.
Watch out that compiler generated temporaries don't increment the reference count. Otherwise you get the poor O(n) vs. O(1) behaviour which I meant weeks ago. (Sorry, I have been busy, so I never read your answer.)
Yes, this is one issue I've been working on.
Matt
