Allow for enabling facet blocks on non-search pages

So, this should basically work the same as the facet browsing blocks - i.e. the query is for everythign in the index (or with e.g. just node access filters applied)?

subscribing

So, I think this needs work to couple it to a specific search environment.

looks like these lines need to be adjusted:

$pages = drupal_strtolower(variable_get('apachesolr_search_facet_pages', ''));

$id = 'apachesolr@' . apachesolr_default_environment();

However, I'm not sure how to avoid overlapping path settings for different environments.

after applying the apachesolr-1252648-6.patch together with the facet api patch (http://drupal.org/files/issues/facetapi-1252644-2.patch for #1252644: Remove assumption that the facets will link back to the page they are on) ,
I got WSOD:

Array
(
    [type] => 1
    [message] => Class name must be a valid object or a string
    [file] => /***/sites/all/modules/ctools/includes/export.inc
    [line] => 1095
)

Then I restored database tables and it is corrected now. Testing further...

why did you change the status?

+++ b/apachesolr_search.moduleundefined
@@ -1255,6 +1293,13 @@ function apachesolr_search_apachesolr_environment_delete($server) {
+  unset($show_facets[$server['env_id']]);
+  unset($facet_pages[$server['env_id']]);

Why does the unset happen here?

Other then this it seems to be ok. Still have to test it though :-)

I'm not sure about performance difference. The variable get is ok but it is an extra thing to take care of when updating or uninstalling. I guess it does make sense to add it to the environment settings. This is cached and is a static variable so it probably does not hurt the performance a lot.

Maybe we should test it with apachesort benchmark? Fire up 1000 requests with variable_get, 1000 with the other one and see how fast it loads?

The results of the benchmark with the patch from cpliakas. Facets were added on a regular page with some bullshit content. No extra modules included.
Will try to change the patch now so it uses the apachesolr_environment_variables and see if it makes significant difference. I suppose we should test it with logged in users also? As mentioned here : http://www.midwesternmac.com/blogs/jeff-geerling/using-apachebench-ab-dr...

notes : no page caching was enabled, query cache for solr was enabled.

Nick-Veenhofs-MacBook-Pro:apachesolr nickveenhof$ ab -c1 -n500 http://apachesolrissues.dev/content/test-utf8
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking apachesolrissues.dev (be patient)
Completed 100 requests
Completed 200 requests
Completed 300 requests
Completed 400 requests
Completed 500 requests
Finished 500 requests

Server Software: Apache/2.2.20
Server Hostname: apachesolrissues.dev
Server Port: 80

Document Path: /content/test-utf8
Document Length: 25000 bytes

Concurrency Level: 1
Time taken for tests: 183.454 seconds
Complete requests: 500
Failed requests: 0
Write errors: 0
Total transferred: 12766500 bytes
HTML transferred: 12500000 bytes
Requests per second: 2.73 [#/sec] (mean)
Time per request: 366.908 [ms] (mean)
Time per request: 366.908 [ms] (mean, across all concurrent requests)
Transfer rate: 67.96 [Kbytes/sec] received

Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.0 0 0
Processing: 327 367 20.3 361 427
Waiting: 304 343 20.1 337 401
Total: 327 367 20.3 361 427

Percentage of the requests served within a certain time (ms)
50% 361
66% 380
75% 386
80% 388
90% 393
95% 400
98% 408
99% 413
100% 427 (longest request)

While I was trying to change this I actually noticed with your patch it is possible to show facets of multiple environments. I guess we should stick to "keep it simple" and only support the default environment for output. It will get very complex I suppose to figure out the correct page where the facet should go to? Working on a patch right now that reduces this complexity

Nick-Veenhofs-MacBook-Pro:facetapi nickveenhof$ ab -c1 -n500 http://apachesolrissues.dev/content/test-utf8
This is ApacheBench, Version 2.3 <$Revision: 655654 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking apachesolrissues.dev (be patient)
Completed 100 requests
Completed 200 requests
Completed 300 requests
Completed 400 requests
Completed 500 requests
Finished 500 requests

Server Software: Apache/2.2.20
Server Hostname: apachesolrissues.dev
Server Port: 80

Document Path: /content/test-utf8
Document Length: 25000 bytes

Concurrency Level: 1
Time taken for tests: 184.370 seconds
Complete requests: 500
Failed requests: 0
Write errors: 0
Total transferred: 12766500 bytes
HTML transferred: 12500000 bytes
Requests per second: 2.71 [#/sec] (mean)
Time per request: 368.741 [ms] (mean)
Time per request: 368.741 [ms] (mean, across all concurrent requests)
Transfer rate: 67.62 [Kbytes/sec] received

Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 0 0.0 0 0
Processing: 326 369 32.2 365 844
Waiting: 302 344 31.5 341 820
Total: 326 369 32.2 365 844

Percentage of the requests served within a certain time (ms)
50% 365
66% 370
75% 373
80% 377
90% 391
95% 406
98% 430
99% 503
100% 844 (longest request)

With attached patch (apachesolr_environment_variable_get())
I also feel this patch is a bit cleaner. It might also impact performance since not every environment is looped now at page load

I just replaced the first benchmark because it was not correct

patch #14 Time per request: 366.908 [ms] (mean)
patch #22 Time per request: 368.741 [ms] (mean)

Mine is a bit slower

I did not really read the full issue, my bad! So I guess for the pages we probably need to load all the environments and loop over those variables if we want to keep it sane?
#8 only mentions that we are coupling it to a specific environment. It did not mention we have to support all the environments at once. My latest patch fetches the default environment and checks the settings of that one. If someone is smart enough he/she would probably make a custom module and make a hook_init for their second or whatever environment and point it to a specific page.

If someone switches from default environment their configurations will also change. At least it is a step in the correct direction I guess. I'm not a big fan of the 'looping over all enviroments approach' because it could end up in an unwanted mess and maybe even in "1 multiplied with the amount of environments" solr queries per page load where the facets are shown.

Committed! This was a great commit before 2012 started!

This doesn't seem to work correctly in a multi-environment setup. It only seems to show the first environments facets, none of the other environments facets show up.