Related Issues
This issue is a sub-task of #1699164: [Meta] Port drupal.org's solr search functionality to D7
- #1549374: Upgrade drupalorg_crosssite to support 6.x-3.x version of Apachesolr (Prep for D7 d.o)
- #1548064: Upgrade drupalorg to support 6.x-3.x version of Apachesolr (Prep for D7 d.o)
Problem/Motivation
Because d.o will transition to D7 before the other *.d.o sites, we need to support multisite apachesolr access across both D6 & D7 sites in the infrastructure.
Multisite search using apachesolr requires 6.x-3.x to work correctly with 7.x sites.
Multisite search using apachesolr also requires apachesolr 6.x-3.x.
So, to facilitate our mixed D6/D7 environment, we'll need to upgrade project to support apachesolr 6.x-3.x and play nicely with apachesolr_multisitesearch 6.x-3.x (and deploy 6.x-3.x versions to all D6 environments in the infrastructure).
Remaining tasks
- getSolrsortUrlQuery isn't a direct port of the previous get_url_queryvalues. Basically, most of the get_url_queryvalues have been ported directly to the getSolrsortUrlQuery which isn't accurate. get_url_queryvalues contained fq params, but getSolrsortUrlQuery does not, so we get (or will get) different behavior for these elements.
Comment | File | Size | Author |
---|---|---|---|
#18 | 1549496-18.patch | 59.87 KB | Nick_vh |
#14 | 1549496-13.patch | 59.76 KB | Nick_vh |
#13 | 1549496-13.patch | 39.86 KB | Nick_vh |
#9 | 1549496-9.patch | 22.02 KB | Nick_vh |
#6 | 1549496-6.patch | 20.49 KB | Nick_vh |
Comments
Comment #1
dwwI'm certainly in favor. However, it's not clear what actually has to happen here. ;) Any chance you could update the summary with more specific info on what needs to be upgraded for this to work, instead of the general motivation for why we want to upgrade?
Thanks,
-Derek
Comment #2
jhedstromI'll update it as I can--the main issue is that the apachesolr api changed names for just about everything between 1.x and 3.x, so mainly what needs to change in Project are function and method names. I have a working patch I'll post soon.
Comment #3
dwwAfter a brief discussion here at the sprint in PDX, we decided the sanest thing to do was to move project_solr into an entirely separate project + Git repo, and then just provide different branches of it that work with apachesolr 6.x-1.x vs. apachesolr 6.x-3.x. Branching all of Project for this seemed like a mess, but just having this small add-on module as a separate project/repo makes a lot of sense.
Comment #4
Nick_vhI'm more than happy to help. Give me an email if you decide to do the sprint. I've been responsible for most of the function and API changes in the solr 6.x-3.x branch (which are exactly the same as 7.x-1.x) so I ow that to drupal.org ;-)
Comment #5
dwwSweet, thanks Nick_vh! I'm going to defer to jhedstrom and csveb10 at this point, since they were the ones moving this forward at the PDX sprint and I've got my hands with a bunch of other efforts happening in parallel. But, if I can be of assistance in any way, please let me know ASAP.
Thanks!
-Derek
Comment #6
Nick_vhFirst attempt,
sorry for the whitespace fixes
Comment #7
csevb10 CreditAttribution: csevb10 commentedNo apologies for fixes of any sort.
I've cleaned up a few issues, and committed the bulk of this patch since it moves us in the right direction and people are sprinting tomorrow, but there is plenty that still needs to be cleaned up.
There are definitely plenty of areas that still need to be addressed, however. Do a search for has_filter to identify some of the problem areas. A lot it centers around the project/modules pages, but some of it affects the surfaced search criteria on the search page as well.
Thanks for the efforts and all the progress that's been made thus far. drupalorg_search & drupalorg_crosssite have been similarly updated.
Comment #8
Nick_vhI'm not even sure if this module needs upgrading at all. I tried drupal.org to see where the paths of this module are in use and it doesn't even seem to be in use?
Comment #9
Nick_vhSo I continued anyway, I figured the hook_menu's were really weird but other stuff was being used. This project does need a lot of love if we want to keep continue it though!
What was done :
Comment #10
Senpai CreditAttribution: Senpai commentedTagging for Sprint 6.
Comment #11
csevb10 CreditAttribution: csevb10 commentedOk, overall this patch (plus a bunch of cleanup on my end) gets us close to a working product.
I see 1 item for cleanup:
Basically, most of the get_url_queryvalues have been ported directly to the getSolrsortUrlQuery which isn't accurate. get_url_queryvalues contained fq params, but getSolrsortUrlQuery does not, so we get (or will get) different behavior for these elements.
We should test that this works for all appropriate use cases:
Once those 2 are taken care of, we should have a generally working project, then, let's see if we can transition the elements out of custom code and better leverage the apachesolr module.
We're close to a working product at this point. Much thanks to nick_vh for all his hard work moving this forward.
Comment #11.0
Senpai CreditAttribution: Senpai commentedAdding sub-tasks
Comment #12
Nick_vhSame here, was this committed to the branch or how does the process work? Please be more clear in what was committed
Comment #13
Nick_vhI was able to work more on this and I think this gets us to a working product. Can you notify me when this gets committed to the 6.x-3.x branch so I can deploy that on the dev server of drupal.org?
Comment #14
Nick_vhMore complete patch here.
This patch fixes the special form for the module facet, also the special page project/$project_type and project/$project_type/categories
Comment #15
Nick_vhComment #16
Nick_vhcheck if this actually works. Should be doable in a less quirky way.
change tid with im_vid_project_vid_id
change tid with im_vid_project_vid_id
change tid with im_vid_project_vid_id
get_url_queryvalues does not exists anymore. Upgrade this
add a newline
decide wether it is better to be consistend and use addField everywhere or addMultipleValue/addValue
remove the im_project_release_api_tids here. Debugging purpose
find a better way to get the facets from the url, leveraging facetapi as the parser.
make sure this stuff is secure. Also leverage facetapi as the parser.
Comment #17
Senpai CreditAttribution: Senpai commentedRetagging from 'ApacheSolr' to 'solr'.
Comment #18
Nick_vhPosting an update
Comment #19
drummCommitted as-is. Review is kinda hard since refactoring is mixed in as well. Followups should be in separate issues.
Comment #20.0
(not verified) CreditAttribution: commentedAdding sub-tasks.