Issue 53359 - Testtool: Add integer type returncode to testtool
Summary: Testtool: Add integer type returncode to testtool
Status: CONFIRMED
Alias: None
Product: utilities
Classification: Unclassified
Component: automation (show other issues)
Version: OOo 1.0.0
Hardware: All All
: P3 Trivial (vote)
Target Milestone: OOo Later
Assignee: AOO issues mailing list
QA Contact: issues@tools
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-08-16 08:02 UTC by joerg.skottke
Modified: 2017-05-20 11:33 UTC (History)
1 user (show)

See Also:
Issue Type: FEATURE
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description joerg.skottke 2005-08-16 08:02:57 UTC
Reason: When scripting the Testtool this might allow a first quick evaluation of
the outcome of a testrun.

Proposal:
rc=0: Successful completion, Testcase finished with no errors/warnings
rc=1: Warnings occurred
rc=2: Errors occurred

Further enhancements:
Depending on the effort the Testtool could return a positive Integer value for
the number of warnings occurred and a negative integer value for the number of
errors occurred where errors take precedence over warnings
Comment 1 gregor.hartmann 2005-08-16 11:34:03 UTC
Is seems that returncodes can only be positive, so the further encancement will
not be possible.
Comment 2 gregor.hartmann 2007-01-24 14:14:32 UTC
How about this:
have a bitfield returned where the least significant bit tells wether there are
errors or warnings and the other bits tell the amount

nnnnnnnx
x = 0 Warnings
x = 1 Errors
nnnnnnn = amount of ...

so if the test runs without warnings and errors the returncode would be 0 (zero)
warnings wuld be even returncodes and errors odd ones
Comment 3 joerg.skottke 2007-01-25 13:20:40 UTC
Yep, that is a possible solution. 
Does this mean that if we have any errors we get the sum of all errors and
warning (errors AFAIK always cause wrnings to pop up as well)?

It would be nicer if we could distinguish the type of error we had in the test -
that is: Error outside of testcase, Error inside a test, warnings and qaerrorlogs.

I guess it is not really important to know exactly how many errors occurred, it
is sufficient to know that there were any at all to find out that the test needs
attention.

So either we could do a OIWQ 4-bit bitfield (Outside errors, inside, warn,
QAErrorLog) or a 16 bit thing with 4 bit for every errorcondition making it
possible to get values up to 15.
Example:

00000001 010101000 = 0 Err. Outside, 1 Err. Inside, 5 warnings and 8 QA Errorlogs.

What do you think?
Comment 4 joerg.skottke 2007-01-25 13:45:17 UTC
Ermh, yes, nice try. I just mixed two entirely different proposals into one
gooey text.

1) Returnvalue does not tell about the amount of errors/warnings, just the types
of errors that occurred.
This would be good enough to tell the tester that he has to review the testcases

2) Returnvalue precisely delivers information about the number of any incident
This would be nice if somebody wants to use the returncode to create statistics.

All three (that includes your proposal) solutions would do while i'd prefer my
number 2 - if this can be implemented with reasonable effort.
Comment 5 Marcus 2017-05-20 11:33:33 UTC
Reset assigne to the default "issues@openoffice.apache.org".