New funit options.

New funit options.

[Chris Moller]

[-- Attachment #1: Type: text/plain, Size: 1001 bytes --]

I've added a couple of options to funit:

 -o, -omit <test-spec>           Specify a test to omit.  Each passed
option will be interpreted as the regex specification of a test to
omit.  This option may be used multiple times.

  -i, -include <test-spec>        Specify a test to include, ovirriding
an omit specification.  Each passed option will be interpreted as the
regex specification of a test to include.  This option may be used
multiple times.


This lets you do stuff like ./funit -o ".*TestExec" to kill tests that
are hanging, or some such naughty behaviour.  Using the -i switch, you
can do stuff like ./funit -o '.*' -i ".*\.sys\..*" to first omit all
tests then selectively include specific ones like, in this case, all the
*.sys.* tests.  Multiple -i and -o switches can be specified--the
effects are cumulative.  Regardless of the order in which -i and -o
switches are specified, the result will be the same: a -i <spec> will
always override a matching -o <spec>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 251 bytes --]

Re: New funit options.

[Andrew Cagney]

Chris these options look great,

In looking over this I noticed a few small nits:
- the code already has .testFiler, this adds a different filtering, can 
the two be merged, instead with just --filter?  Java regular expressions 
let you specify exclusions directly
- or, instead of -o (which typically is associated with output) is 
--exclude better?
- for coding, again, please use (foo == null) and not the reverse; in 
fact for that code it is simpler/easier to just always create the array 
and then let the for() loop using .size() as the upper bound
- was the man page updated

Andrew

Chris Moller wrote:
> I've added a couple of options to funit:
>
>  -o, -omit <test-spec>           Specify a test to omit.  Each passed
> option will be interpreted as the regex specification of a test to
> omit.  This option may be used multiple times.
>
>   -i, -include <test-spec>        Specify a test to include, ovirriding
> an omit specification.  Each passed option will be interpreted as the
> regex specification of a test to include.  This option may be used
> multiple times.
>
>
> This lets you do stuff like ./funit -o ".*TestExec" to kill tests that
> are hanging, or some such naughty behaviour.  Using the -i switch, you
> can do stuff like ./funit -o '.*' -i ".*\.sys\..*" to first omit all
> tests then selectively include specific ones like, in this case, all the
> *.sys.* tests.  Multiple -i and -o switches can be specified--the
> effects are cumulative.  Regardless of the order in which -i and -o
> switches are specified, the result will be the same: a -i <spec> will
> always override a matching -o <spec>
>
>   

Re: New funit options.

[Chris Moller]

[-- Attachment #1: Type: text/plain, Size: 2345 bytes --]



Andrew Cagney wrote:
> Chris these options look great,
>
> In looking over this I noticed a few small nits:
> - the code already has .testFiler, this adds a different filtering,
> can the two be merged, instead with just --filter?  

The options could easily be merged, of course, but I left the original
alone because I don't like mucking up Other People's Code.  The thing to
do might be to have the -all and -stress flags just affect the new
includeTests and omitTests String[]s and get rid of the older testFilter
String.

> Java regular expressions let you specify exclusions directly

I didn't know that (a statement that globally applies to me in all
things Java).  I'll go RTFM and patch accordingly.

> - or, instead of -o (which typically is associated with output) is
> --exclude better?

Done.

> - for coding, again, please use (foo == null) and not the reverse; in
> fact for that code it is simpler/easier to just always create the
> array and then let the for() loop using .size() as the upper bound

Okay.  (Very old habits--I wrote my first line of code well over a third
of a century ago, back when memory was expensive.)

> - was the man page updated

Nope, forgot.  Will do.

>
> Andrew
>
> Chris Moller wrote:
>> I've added a couple of options to funit:
>>
>>  -o, -omit <test-spec>           Specify a test to omit.  Each passed
>> option will be interpreted as the regex specification of a test to
>> omit.  This option may be used multiple times.
>>
>>   -i, -include <test-spec>        Specify a test to include, ovirriding
>> an omit specification.  Each passed option will be interpreted as the
>> regex specification of a test to include.  This option may be used
>> multiple times.
>>
>>
>> This lets you do stuff like ./funit -o ".*TestExec" to kill tests that
>> are hanging, or some such naughty behaviour.  Using the -i switch, you
>> can do stuff like ./funit -o '.*' -i ".*\.sys\..*" to first omit all
>> tests then selectively include specific ones like, in this case, all the
>> *.sys.* tests.  Multiple -i and -o switches can be specified--the
>> effects are cumulative.  Regardless of the order in which -i and -o
>> switches are specified, the result will be the same: a -i <spec> will
>> always override a matching -o <spec>
>>
>>   
>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 251 bytes --]

Re: New funit options.

[Chris Moller]

[-- Attachment #1: Type: text/plain, Size: 1857 bytes --]

A few minor things just committed.

Andrew Cagney wrote:
> Chris these options look great,
>
> In looking over this I noticed a few small nits:
> - the code already has .testFiler, this adds a different filtering,
> can the two be merged, instead with just --filter?  Java regular
> expressions let you specify exclusions directly

This will take more time--deferred.

> - or, instead of -o (which typically is associated with output) is
> --exclude better?

Done.

> - for coding, again, please use (foo == null) and not the reverse; in
> fact for that code it is simpler/easier to just always create the
> array and then let the for() loop using .size() as the upper bound

Done.

> - was the man page updated

Done.

>
> Andrew
>
> Chris Moller wrote:
>> I've added a couple of options to funit:
>>
>>  -o, -omit <test-spec>           Specify a test to omit.  Each passed
>> option will be interpreted as the regex specification of a test to
>> omit.  This option may be used multiple times.
>>
>>   -i, -include <test-spec>        Specify a test to include, ovirriding
>> an omit specification.  Each passed option will be interpreted as the
>> regex specification of a test to include.  This option may be used
>> multiple times.
>>
>>
>> This lets you do stuff like ./funit -o ".*TestExec" to kill tests that
>> are hanging, or some such naughty behaviour.  Using the -i switch, you
>> can do stuff like ./funit -o '.*' -i ".*\.sys\..*" to first omit all
>> tests then selectively include specific ones like, in this case, all the
>> *.sys.* tests.  Multiple -i and -o switches can be specified--the
>> effects are cumulative.  Regardless of the order in which -i and -o
>> switches are specified, the result will be the same: a -i <spec> will
>> always override a matching -o <spec>
>>
>>   
>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 251 bytes --]