You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by Alexandre Rafalovitch <ar...@gmail.com> on 2013/03/13 01:32:09 UTC
copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Hello,
I have an example schema which worked in 4.1 but is failing to load in 4.2
with: "copyField source :'addr_*' is not an explicit field and doesn't
match a dynamicField".
I think this must be due to SOLR-3798, but I don't understand why even
after reading it through several times.
My schema (excerpt) is:
<field name="addr_from" type="email" indexed="true" stored="true"
required="true" />
<field name="addr_to" type="email" multiValued="true" indexed="true"
stored="true" required="true" />
<copyField source="addr_*" dest="text" />
I thought this would have been a valid use case. Can someone with deeper
understanding of this aspect explain what I am missing.
Regards,
Alex.
Personal blog: http://blog.outerthoughts.com/
LinkedIn: http://www.linkedin.com/in/alexandrerafalovitch
- Time is the quality of nature that keeps events from happening all at
once. Lately, it doesn't seem to be working. (Anonymous - via GTD book)
Re: copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Posted by Steve Rowe <sa...@gmail.com>.
I committed a fix under SOLR-4567.
On Mar 13, 2013, at 12:50 AM, Steve Rowe <sa...@gmail.com> wrote:
> Yes, this is a regression, definitely my fault. Sorry Alex!
>
> The table on SOLR-3798 is missing this case: a glob matching one or more explicit fields (as opposed to dynamic fields).
>
> I've filed a JIRA: https://issues.apache.org/jira/browse/SOLR-4567
>
> On Mar 13, 2013, at 12:20 AM, "Jack Krupansky" <ja...@basetechnology.com> wrote:
>> And, the wiki does not note the decommissioning of a useful feature of copyField. Although, the wiki is woefully incomplete when it comes to glob patterns for fields.
>
> I agree - I wrote what I thought would be a good addition to the wiki just above the copyField combinations table on SOLR-3798. But it needs additional verbiage to cover Alex's case.
>
>> Reading the table in SOLR-3798 as carefully as I can, it seems to indicate that your use case is supposed to be supported as case #9, leading me to conclude that it may simply be a bug that your use case is failing in 4.2.
>>
>> 9 subset pattern <field> name <copyField source="*_src_sub_i" dest="title"/> Yes Yes
>
> Alex's case is different from what I call "subset patterns" on SOLR-3798, since that's shorthand for "subset of the language accepted by the pattern for a referenced dynamic field".
>
> I'll make a copy of that table and add a case where the source value type can be a glob matching one or more explicit fields.
>
> Steve
>
>
Re: copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Posted by Jack Krupansky <ja...@basetechnology.com>.
Thanks for the clarification! Although, maybe we need to come up with some
simpler, more clear terminology.
-- Jack Krupansky
-----Original Message-----
From: Steve Rowe
Sent: Wednesday, March 13, 2013 12:50 AM
To: solr-user@lucene.apache.org
Subject: Re: copyField with * stops working with 4.2 (related to SOLR-3798
?)
Yes, this is a regression, definitely my fault. Sorry Alex!
The table on SOLR-3798 is missing this case: a glob matching one or more
explicit fields (as opposed to dynamic fields).
I've filed a JIRA: https://issues.apache.org/jira/browse/SOLR-4567
On Mar 13, 2013, at 12:20 AM, "Jack Krupansky" <ja...@basetechnology.com>
wrote:
> And, the wiki does not note the decommissioning of a useful feature of
> copyField. Although, the wiki is woefully incomplete when it comes to glob
> patterns for fields.
I agree - I wrote what I thought would be a good addition to the wiki just
above the copyField combinations table on SOLR-3798. But it needs
additional verbiage to cover Alex's case.
> Reading the table in SOLR-3798 as carefully as I can, it seems to indicate
> that your use case is supposed to be supported as case #9, leading me to
> conclude that it may simply be a bug that your use case is failing in 4.2.
>
> 9 subset pattern <field> name <copyField source="*_src_sub_i"
> dest="title"/> Yes Yes
Alex's case is different from what I call "subset patterns" on SOLR-3798,
since that's shorthand for "subset of the language accepted by the pattern
for a referenced dynamic field".
I'll make a copy of that table and add a case where the source value type
can be a glob matching one or more explicit fields.
Steve
Re: copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Posted by Steve Rowe <sa...@gmail.com>.
Yes, this is a regression, definitely my fault. Sorry Alex!
The table on SOLR-3798 is missing this case: a glob matching one or more explicit fields (as opposed to dynamic fields).
I've filed a JIRA: https://issues.apache.org/jira/browse/SOLR-4567
On Mar 13, 2013, at 12:20 AM, "Jack Krupansky" <ja...@basetechnology.com> wrote:
> And, the wiki does not note the decommissioning of a useful feature of copyField. Although, the wiki is woefully incomplete when it comes to glob patterns for fields.
I agree - I wrote what I thought would be a good addition to the wiki just above the copyField combinations table on SOLR-3798. But it needs additional verbiage to cover Alex's case.
> Reading the table in SOLR-3798 as carefully as I can, it seems to indicate that your use case is supposed to be supported as case #9, leading me to conclude that it may simply be a bug that your use case is failing in 4.2.
>
> 9 subset pattern <field> name <copyField source="*_src_sub_i" dest="title"/> Yes Yes
Alex's case is different from what I call "subset patterns" on SOLR-3798, since that's shorthand for "subset of the language accepted by the pattern for a referenced dynamic field".
I'll make a copy of that table and add a case where the source value type can be a glob matching one or more explicit fields.
Steve
Re: copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Posted by Jack Krupansky <ja...@basetechnology.com>.
Solr-4503 made the changes to copyField semantics. Indeed, it is not clear
whether Solr-4503 (or even Solr-3798) was really intended to de-commit
existing functionality. I mean, the normal procedure is to deprecate a
feature long before removing it.
And, the wiki does not note the decommissioning of a useful feature of
copyField. Although, the wiki is woefully incomplete when it comes to glob
patterns for fields.
Reading the table in SOLR-3798 as carefully as I can, it seems to indicate
that your use case is supposed to be supported as case #9, leading me to
conclude that it may simply be a bug that your use case is failing in 4.2.
9 subset pattern <field> name <copyField source="*_src_sub_i"
dest="title"/> Yes Yes
So, I'd go ahead and file this as a bug.
Steve?
https://issues.apache.org/jira/browse/SOLR-3798
https://issues.apache.org/jira/browse/SOLR-4503
The revision that made the change:
http://svn.apache.org/viewvc?view=revision&revision=1453162
-- Jack Krupansky
-----Original Message-----
From: Alexandre Rafalovitch
Sent: Tuesday, March 12, 2013 8:32 PM
To: solr-user@lucene.apache.org
Subject: copyField with * stops working with 4.2 (related to SOLR-3798 ?)
Hello,
I have an example schema which worked in 4.1 but is failing to load in 4.2
with: "copyField source :'addr_*' is not an explicit field and doesn't
match a dynamicField".
I think this must be due to SOLR-3798, but I don't understand why even
after reading it through several times.
My schema (excerpt) is:
<field name="addr_from" type="email" indexed="true" stored="true"
required="true" />
<field name="addr_to" type="email" multiValued="true" indexed="true"
stored="true" required="true" />
<copyField source="addr_*" dest="text" />
I thought this would have been a valid use case. Can someone with deeper
understanding of this aspect explain what I am missing.
Regards,
Alex.
Personal blog: http://blog.outerthoughts.com/
LinkedIn: http://www.linkedin.com/in/alexandrerafalovitch
- Time is the quality of nature that keeps events from happening all at
once. Lately, it doesn't seem to be working. (Anonymous - via GTD book)