Hello Shawn,
Ah, forgot about that thread. But yes, it makes sense and we used it as such in
production. But the test was never updated, never failed, until 7.4.0 came
around. So it is confusing why the tests didn't fail, and confusing why it does
in 7.4.0. Any ideas why now?
Thanks,
Markus
-----Original message-----
> From:Shawn Heisey <apa...@elyograg.org>
> Sent: Wednesday 27th June 2018 17:40
> To: solr-user@lucene.apache.org
> Subject: Re: 7.4.0 changes in DocTransformer behaviour
>
> On 6/27/2018 8:29 AM, Markus Jelsma wrote:
> > I am attempting an upgrade to 7.4.0, besides a decent amount of thread leak
> > warnings i suddenly get, there are some strange failing tests, of which one
> > is one of four DocTransformer plugins. That plugin is really simple, it
> > just transformers integer or long numerics into a binary string of 0 and 1.
> >
> > This is the test:
> >
> > assertQ( req("q", "*:*", "indent", "true", "fl",
> > "id,minhash:[binstr]"),
> > "//*[@numFound='" + max + "']",
> >
> > "//str[.='0000000000000000000000000000000000000000000000000000000000000000']",
> >
> > "//str[.='0000000000000000000000000000000000000000000000000000000000000001']",
> >
> > "//str[.='0000000000000000000000000000000000000000000000000000000000000010']",
> >
> > "//str[.='0000000000000000000000000000000000000000000000000000000000000011']");
> > }
> >
> > Changing the fl parameter into "*,minhash:[binstr]" solves the problem, the
> > test passes.
>
> I will admit that I do not know exactly what that assert syntax means,
> but I'm guessing that it looks for those strings of numbers in the
> response, and makes sure that numFound is what you expect it to be.
>
> This problem sounds exactly the same as the problem you were
> encountering and asked the list about in December 2016, on version 6.3.0
> - the transformed field you've requested with fl doesn't work unless you
> also request all fields. Here's that thread on Nabble:
>
> http://lucene.472066.n3.nabble.com/DocTransformer-not-always-working-td4309785.html
>
> I do not really understand what Hoss told you in response to that thread
> ... but can you use the same information this time?
>
> Thanks,
> Shawn
>
>
