Greetings Farid.
The reason for the 'DATA' on windows is because the locales aren't being
tested at this time. The reason they aren't being tested is because the
glue script hasn't been designed to parse the test driver output,
generated by the locale test script. Under the Unix infrastructure, the
locale test scripts are run using the exec utility. If consistency
between the Unix and Windows infrastructure is desired, it would make
sense to run the locale tests using the exec utility.
We actually do capture the HTML output within our automated testing
system, but it isn't part of the published nightly results, as it would
be a special case. The build log (the first file associated with each
build) is the output from stdout/stderr when running the glue scripts.
One enhancement I would appreciate seeing would be the output of the
library and example build process on the standard output from the
build_(compiler).bat script, as it provides no indication of the
progress being made while the script runs.
--Andrew Black
Farid Zaripov wrote:
-----Original Message-----
From: Martin Sebor [mailto:[EMAIL PROTECTED]
Sent: Monday, November 13, 2006 6:53 PM
To: [email protected]
Subject: Re: RFC: publishing nightly test results
Any comments or questions on this? Farid?
Not yet. I found it useful as for me.
I see "DATA" in State column in rows with results on MSVC. The tooltop
comment says
"Incomplete data". Can I help with this? The windows build
infrastructure produces the build
log (i.e. msvc-7.1-15s.html) and perhaps this log can be attached? The
size of this log file
is big due to HTML format and UNICODE characters (on MSVC8). We can
modify
makelog.wsf script to produce the ASCII text log file as well as html
one.
Farid.
