|
From: Ashley P. <api...@co...> - 2008-02-29 15:54:48
|
I've been looking at my output aggregation tool again this week and have spotted a problem with the new output qualifier, in the old system with the --log-file-qualifier option the name and value of the variable are put into the xml file, under the new system this information is no longer available so my tool isn't able to sort the output from multiple files. An example of the old output is shown below, does anybody have any comments on this before I work up a patch to re-instate the old behaviour? <logfilequalifier> <var>SCORE_SELF_NODE</var> <value>0</value> </logfilequalifier> Ashley, |
|
From: Julian S. <js...@ac...> - 2008-03-01 20:19:10
|
On Friday 29 February 2008 16:54, Ashley Pittman wrote:
> I've been looking at my output aggregation tool again this week and have
> spotted a problem with the new output qualifier, in the old system with
> the --log-file-qualifier option the name and value of the variable are
You're absolutely right. Sorry for the breakage.
I guess one way to fix it would be to scan the --log-file string
(which should be pointed to by VG_(clo_log_name)).
When you see ${FOO} in the string, emit
<logfilequalifier> <var>FOO</var> <value>$FOO</value>
</logfilequalifier>
where obviously $FOO denotes the value of FOO at the time you
process the string.
You can find some code used to deal with ${FOO} in VG_(expand_file_name).
Perhaps that could be factored into a function used by both the
current log-file-name handling, and the logfilequalifier stuff.
J
|
|
From: Ashley P. <api...@co...> - 2008-03-03 16:56:22
Attachments:
xml-diff.patch
|
On Sat, 2008-03-01 at 21:15 +0100, Julian Seward wrote:
> On Friday 29 February 2008 16:54, Ashley Pittman wrote:
> > I've been looking at my output aggregation tool again this week and have
> > spotted a problem with the new output qualifier, in the old system with
> > the --log-file-qualifier option the name and value of the variable are
>
> You're absolutely right. Sorry for the breakage.
It's as much my fault as yours, I didn't spot it either.
> You can find some code used to deal with ${FOO} in VG_(expand_file_name).
> Perhaps that could be factored into a function used by both the
> current log-file-name handling, and the logfilequalifier stuff.
I've simply copied the function to m_main.c and stripped out the bits I
don't need.
Ashley,
|