jmeter-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <seb...@gmail.com>
Subject Re: regex for jmeter.reportgenerator.exporter.html.series_filter
Date Fri, 06 May 2016 09:21:31 GMT
On 6 May 2016 at 09:44, Philippe Mouawad <philippe.mouawad@gmail.com> wrote:
> On Friday, May 6, 2016, sebb <sebbaz@gmail.com> wrote:
>
>> On 5 May 2016 at 19:52, Philippe Mouawad <philippe.mouawad@gmail.com
>> <javascript:;>> wrote:
>> > On Thu, May 5, 2016 at 8:29 PM, Felix Schumacher <
>> > felix.schumacher@internetallee.de <javascript:;>> wrote:
>> >
>> >>
>> >>
>> >> Am 5. Mai 2016 20:17:37 MESZ, schrieb Philippe Mouawad <
>> >> philippe.mouawad@gmail.com <javascript:;>>:
>> >> >Hi Felix,
>> >> >
>> >> >1. true
>> >> >2. I don't understand the problem , can you clarify for me ?
>> >>
>> >> ^ indicates start of line (String in our case). But the code does a
>> match,
>> >> so that is implicit. But if I see a ^ I think the regex will do a "find"
>> >> instead of a "match".
>> >>
>> >> So as a result I think we should remove the ^.
>> >>
>> >> >
>> >> >3. Yes we want to trap:
>> >> >- Sample1
>> >> >- Sample1-success
>> >> >- Sample1-failure
>> >> >- Sample2
>> >> >- Sample2-success
>> >> >- Sample2-failure
>> >> >
>> >> >Ideally not Sample2XXX-success.
>> >>
>> >> And not Sample123?
>> >>
>> > No
>> >
>> >>
>> >> >Do you see a better regex ?
>> >>
>> >> If we want to match all those, than the regex is correct.
>> >>
>> >> All in all, I think it would be safe to change the regex to
>> >>
>> >> Sample[12](?:-failure|-success)?
>> >>
>> >
>> > Yes but in real life they will not be named Sample1, Sample2 , but
>> > Purchase, Search for example
>>
>> In which case we seem to be asking the customer to provide their own regex.
>
> yes
>
>>
>> This is prone to error, and won't they have to change it for each test?
>
>
> I don't understand.

I mean that creating regexes can be hard.

> Samples naming is up to users, it's already like this in BackendListener,
> same for regexp extractor.
>
> Here we just provide an example of a regexp with 2 samples.

In which case use names which are more representative, rather than
ones which happen to be easy to write regexes for.

>
>>
>> I think this may cause a lot of questions on the user list.
>>
>> If it is just a question of matching some words optionally followed by
>> '-success' or '-failure', then I think the user should just have to
>> supply the words.
>> Possibly even as lines in a separate file.
>> The code then creates its own regex or does its own matching.
>
>
> Regexp is more flexible and there are a lot of helper on the net.

But we still get quite a few questions on them.

> To be coherent with BackendListener we should stick with regexp.

If you are referring to the "samplersList" variable in BackendListener
then using it as a regex is *optional*

Also the field is one the GUI, so each test can have its own config
(or even multiple ones in a single test)

> A list of samples in another file is again another configuration to manage
> instead of having everything in user.properties.

But if the regex need to be changed for each test then having it in
^[jmeter|user]\.properties$ is not very convenient.

I've also just noticed that the property appears in both
jmeter.properties and user.properties.

That's more potential confusion.

I think the whole way that ReportGenerator uses properties needs to be
looked at; I'll raise a separate thread.

> Let's see what happens with 3.0 and we can enhance in the future.
> If you look at it we already had feedback on reports, and it was not one of
> them.
>
>
>>
>> >>
>> >> Or, if it should be a bit more descriptive
>> >>
>> >> (?:Sample1|Sample2)(?:-failure|-success)?
>> >>
>> >> Which is mostly the same as the one in the properties file. Differences
>> >> are non capture and remove indicator of start of line.
>> >>
>> > ok, can you just double check it ?
>> >
>> >>
>> >> Felix
>> >>
>> >> >Thanks
>> >> >
>> >> >
>> >> >
>> >> >On Thu, May 5, 2016 at 7:23 PM, Felix Schumacher <
>> >> >felix.schumacher@internetallee.de <javascript:;>> wrote:
>> >> >
>> >> >> Hi all,
>> >> >>
>> >> >> the regex for series_filter is currently set to
>> >> >>
>> >> >> ((^Sample1)|(^Sample2))(-success|-failure)?
>> >> >>
>> >> >> in the user.properties file.
>> >> >>
>> >> >> The regex could be written a bit shorter as
>> >> >>
>> >> >> ^Sample[12](-success|-failure)?
>> >> >>
>> >> >> But there are still a few things to consider.
>> >> >>
>> >> >> 1. I don't think that we are interested in the captured groups
and
>> >> >> could tell the regex engine that by using (?:...) instead of (...).
>> >> >>
>> >> >> 2. The ^ in front of Sample1 makes it look like the regex would
be
>> >> >used
>> >> >> as "find", as there is no $ to indicate the end of a line.
>> >> >>
>> >> >> 3. The ? after the last group indicates that the results could
be one
>> >> >> of
>> >> >>   + Sample1
>> >> >>   + Sample1-success
>> >> >>   + Sample1-failure
>> >> >>   + Sample2...
>> >> >>   Is this true? Or is just ...-success and ...-failure? In that
case
>> >> >> the ? at the end of the regex should be removed.
>> >> >>
>> >> >> Regards,
>> >> >>  Felix
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >>
>> >>
>> >
>> >
>> > --
>> > Cordialement.
>> > Philippe Mouawad.
>>
>
>
> --
> Cordialement.
> Philippe Mouawad.

Mime
View raw message