Boost Testing :
From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2007-02-16 00:53:35
Roland Schwarz writes:
> The file expected_results.xml looks like being important for what gets
> displayed on the regression tables.
Yes, it is.
> And so I think it is also relevant for making progress with
> regression testing towards a release. (This is why I CC'ed Thomas
> I was unable to find any documentation about this file or it's purpose.
It holds the regression results from the "last known good" release
(usually the last official release).
> My attention was triggered because this file seems to be responsible how
> toolset tagging works. (i.e. aliasing works)
> My questions:
> 0) Is there somewhere a description available?
The above is basically all there is to it. WRT 1.34, you might find
this message to be of a help:
> 1) Is this file supposed to be edited by means of a plain text editor,
> e.g. vi?
In a number of limited cases, yes.
> 2) If this file is supposed to be edited by a plain text editor, may I
> please check in a version of the file that has suitable enough line-
> breaks, so that it can be edited, and does not look like a long
> string of garbage?
If you can guarantee that this will not affect its structural content,
please go ahead.
> 3) After having reformated the file I discovered, that there is a _lot_
> of double entries. e.g.
> Wouldn't it be meaningful to remove these doublets, so the result
> rather will read:
> <toolset name="mingw-3_3_1">
> <library name="algorithm/minmax">
> <test-result test-name="minmax" result="success"/>
> <test-result test-name="minmax_element" result="success"/>
It would make it more manageable size-wise, but I wouldn't spend your
time on it (unless you plan to do it programmatically):
- The file is huge and the duplicates are harmless (besides
contributing to its size).
- The file is automatically generated using
and any manual effort will be wasted after it's regenerated.
-- Aleksey Gurtovoy MetaCommunications Engineering