[pytest-dev] junitxml incremental output writing

[Ronny Pfannschmidt]

Hi Floris,

this morning i remebered another idea.

instead of creating the junitxml report direcly,
one could use the new reportlog plugin, and add report replay to it, 
thus collecting the report logs on the first run, then combiningthe 
pytest reports to junitxml on the second run.


-- Ronny

Am 06.04.20 um 21:45 schrieb Floris Bruynooghe:
> Hey Ronny,
>
> On Mon 06 Apr 2020 at 09:24 +0200, Ronny Pfannschmidt wrote:
>> an incremental writer has to continuously update either the
>> pre-overstretched xml header of the report or continuously rewrite the
>> file at once.
> Cool. Thanks for putting this so succinctly!
>
>> Another approach could be to just write the xml report parts of each
>> item into a junitxml.parts file and
>> at the end of the test to just recombine the file with the enclosing xml
>> element.
> This is possibly an interesting approach.  I think CI can usually
> already combine many reports into one so the simplest implementation
> could be to write one report per test and not even bother with
> recombining.
>
>> I think people that care about it should come up with it.
> Here I agree entirely.  I'm only curious to explore the general ideas a
> bit so I know what to tell people to work on when they ask about this.
>
>
> No one feel like critiquing the idea of providing an API to plugins that
> can be used to trigger reports to be (re)written right now?
>
>
> Cheers,
> Floris