[Radiant] Something strange here
Was working with a development environment when suddenly the admin pages started looking funny, and I'm getting some javascript errors. The js errors are centered around Behavior, which comes from lowpro.js, and I'm noticing that the admin pages aren't loading it correctly. not sure what I screwed up, but when I try to go back and rebuild from fresh, I'm getting undefined method 'locale' after logging in. Any idea what I screwed up? Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Something strange here
On Apr 1, 2010, at 4:35 PM, john muhl wrote: On Thu, Apr 1, 2010 at 3:14 PM, Arlen Walker arlen.wal...@gmail.com wrote: Was working with a development environment when suddenly the admin pages started looking funny, and I'm getting some javascript errors. The js errors are centered around Behavior, which comes from lowpro.js, and I'm noticing that the admin pages aren't loading it correctly. not sure what I screwed up, but when I try to go back and what does the error console have to say about the javascript errors? TypeError: Result of expression 'Behavior' [undefined] is not an object. Far as I can tell, this is because lowpro.js is not getting loaded after prototype and before the rest of the js files rebuild from fresh, I'm getting undefined method 'locale' after logging in. this part sounds like you were running an edge version and then updated to a newer edge version and didn't run `rake db:migrate` Yep, you and Jim both nailed this one. Usually when I do a fresh install I do :update before doing :bootstrap, it appears either this time I didn't or the updated migration isn't in bootstrap, because the migration fixed that. I'm beginning to suspect the js errors are coming from an installation that's an unholy hybrid between 0.8.1 and 0.9. IIRC, it was created by the radiant-rc gem (there seems to be no way of having both the rc *and* the 0.8.1 gem on your system at the same time -- at least the radiant executable identifies itself as 0.9 after installing radiant-rc, and I don't find the 0.8.1 executable anywhere on the system anymore) but I used rake radiant:freeze:edge TAG=0.8.1 to freeze it, so it used the current stable code. Does that sound likely as a cause? Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues
BTW, if any are curious, here's what I think I know about this (sorry, forgot to report back about the experiments): rake spec or rake spec:models load *only* the base radiant db migrations into the test database -- they load no extension migrations at all. This means that if any extension modifies the base tables in its migrations, those migrations are not loaded during this test. The only time any extension migrations get loaded into the test db is under rake spec:extensions when they do get loaded, when the tests are run for that particular extension. (This may not be entirely correct, as I realized as I wrote the above I didn't have any integration tests in my local tree involving reorder, and those might have caused the reorder migrations to load.) Since the reorder extension modifies children_find_options (replacing 'published_at' with 'position' as the default) it breaks the tests, as the position column migration has not been loaded when the model tests are being run. If it had simply created a position column and not altered default behavior (merely giving you the additional option of using by=position') the errors would not have appeared, and my confusion would have been less. BTW, am I right in assuming there isn't a command to test a specific extension, only one which tests all installed extensions at once? Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues (Final Solution)
On Mar 30, 2010, at 12:33 PM, john muhl wrote: On Tue, Mar 30, 2010 at 9:53 AM, Arlen Walker arlen.wal...@gmail.com wrote: BTW, am I right in assuming there isn't a command to test a specific extension, only one which tests all installed extensions at once? rake spec:extensions EXT=ext_name Ah, thank you. BTW, have a solution to this issue, finally. Radiant is not updating the schema.rb file when extensions are migrated. After doing an extension migration use: rake db:schema:dump to update the schema.rb file. Then the test db will be recreated properly during testing. So much simpler when I finally realized 'rake spec' was using schema:load and not remigrating everything every time. (What can I say? I've been sick so my mind has been slower than usual.) Radiant needs to do a schema:dump after the extension migrations happen, but until then, doing it manually seems to work. Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues
On Fri, Mar 19, 2010 at 3:38 PM, Jim Gay j...@saturnflyer.com wrote: Don't try to fix it. If a core spec is defined to expect things in a default order but an extension changes the default, then those added specs will fail. The only way around this would be to undefine the original specs, but I'm not sure of a way to do that. I'll add a +1 to Jim on that. The right way to handle this would be for the new find method to test and and see if an order clause was passed in, and use that clause if it was, while defaulting to position order if no other order is specified. (If the core spec insists on another default order, *then* it's time to address the idea of changing the core spec, as that approach is almost guaranteed to cause issues with extensions.) I've had lots of issues with interactions between both drag_order and reorder and r:children tags because of that. (For some items, I want descending order by date, for others I want a specific page order.) I'd planned on forking reorder to do that (I'm migrating a wordpress site with 150+ posts, and I'm sick of reordering them as I migrate them; the current state of reorder lets the asc and desc values through, but overrides the date_published order with position order) once I got the tests to run sensibly. I'm having trouble with that at the moment, though. Which means that the column doesn't exist. So it sounds to me like Anna is merely seeing the result of the altered behavior with a position column, but Arlen doesn't even have the position column. Yep. I can run the migration and look at the test db and see the position column, but when I run the tests, the test db gets rebuilt *without* the position column, which I've verified. I'm still sure I'm overthinking this, and when I find out what's wrong I'm going to feel stupid, but so far I'm still mystified. Gonna put it down for a while and come back to it with fresh eyes. I haven't had a chance to really look into this, but I have seen the problem before and can't recall how it was fixed for me (other than running rake RAILS_ENV=test db:migrate:extensions) That actually worked in one installation here, the 0.8.1 one. But it just stares at me blankly in the edge one. I'm sure I've screwed something up, and it's something simple. It just feels like it. (I'm liking the edge code better than 0.8.1 at the moment, so I'll probably be using it for the migration.) ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
[Radiant] Reorder Extension testing issues
This is actually common to several extensions, so I'm fairly sure I screwed up something on my end, but it's just to obvious for me to see it. Starting from a completely fresh installation: 1) radiant radiant -d mysql 2) cd radiant 3) rake radiant:freeze:edge TAG=0.8.1 4) cd vendor/radiant At this point I edit the database.yml file at radiant/vendor/radiant/config/database.yml 5) rake db:bootstrap (Don't know if it matters but the last time I did this I chose the Roasters template) 6) rake spec:models The last line was to verify that the installation into the directory called 'radiant' seemed to go properly, and it does appear to have, since all tests passed with a deprecation warning. So now I proceed with downloading the reorder extension, unzipping it, renaming the resulting directory to reorder and dropping it in radiant/vendor/radiant/vendor/extensions (perhaps I should have chosen a less confusing project name, but I didn't expect to have this issue when I chose it). From that point: 7) rake development radiant:extensions:reorder:migrate 8) rake development radiant:extensions:reorder:update (I'm running apache2 with Passenger, and have set the default RailsEnv to development.) 9) rake spec:models and 66 of the tests that passed the first time fail this time. All of the failures are along the lines of: Mysql::Error: Unknown column 'position' in 'order clause': so it seems as if the migration that has occurred in line 7 above (verified by inspecting the db) never happens during the testing. The extension itself appears to be functioning fine in the application, as the position column is filled in correctly, and it does move the pages up and down, so it looks as if only the testing environment is failing to apply the migration for some reason. This sort of thing is obvious enough that I'm sure I'm just missing something simple, but I'm hanged if I can see it. As for variations, I've tried this with both radiant and radiant-rc, and I've frozen to edge and updated as well as freezing to 0.8.1, with no difference. Highlights from the local gem list are: cucumber (0.6.3, 0.3.104) cucumber-rails (0.3.0) passenger (2.2.11, 2.2.10, 2.2.9) radiant (0.8.1, 0.8.0) rails (2.3.5, 2.2.2, 1.2.6) rake (0.8.7, 0.8.3) rspec (1.3.0, 1.2.6) rspec-rails (1.3.2, 1.2.6) rubygems-update (1.3.6, 1.3.5) webrat (0.7.0, 0.4.4) ZenTest (4.2.1) Didn't want to post the whole list, let me know if there's another one that might be relevant. Ruby is 1.8.7 patchlevel 72, default install on SnowLeopard. (Yes, I know. I keep meaning to upgrade to 1.9.x but since this isn't a public-facing machine the urgency just isn't there, and I don't figure it's worth the hassle to upgrade to anything less than 1.9.x.) Ideas? Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues
On Mar 18, 2010, at 5:23 PM, banane wrote: You have to rake migrate the test env: rake radiant:extensions:reorder:migrate RAILS_ENV=test Yet when I do that, it says: Missing these required gems: rspec-rails ~ 1.2.6 Yet gem list rspec-rails --local shows it's installed: *** LOCAL GEMS *** rspec-rails (1.3.2, 1.2.6) (The 1.3.x versions of rspec and rspec-rails came along with the cucumber 0.6.3 update I installed.) Even so, when I removed both versions of rspec-rails and reinstalled only the 1.2.6 version, the migrate happened (verified by inspecting the test db for the column, and the 'reorder-1' in the schema migrations table): $ rake radiant:extensions:reorder:migrate RAILS_ENV=test (in /Users/arlenwalker/Sites/radiant/vendor/radiant) == AddPositionToPages: migrating = -- add_column(:pages, :position, :integer) - 0.3256s -- Putting all pages in a default order... - 0.2469s == AddPositionToPages: migrated (0.5728s) However, it made no difference, as after rake spec:models the same 66 tests failed. Interestingly enough, after running the tests, the position column was again dropped, and the extension's migration not applied during testing. I can also supply the info that some other extensions that create tables for their own use do not have the tables created, but I figured if I can figure out what's wrong here, the solution will apply elsewhere. Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues
On Mar 18, 2010, at 5:23 PM, banane wrote: Or it might work with just a migration all together: rake db:migrate RAILS_ENV=test Actually, oddly enough, a variation of this changed the results: Instead of being at radiant/vendor/radiant when I did the migration, I dropped back to the project working directory itself (radiant, in this case) did the migration and then rake spec:models and now 20 tests failed, and none of the failures relate to position, and the position column remained in the test table. (moving back to vendor/radiant and rerunning the tests repeated the 66 failures). So doing the full migration gave different results from the extension migrate and I suspect some of them at least come from position overriding expected behavior (such as a couple that complained that using the 'order' attribute didn't change the order. Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
Re: [Radiant] Reorder Extension testing issues
On Mar 18, 2010, at 7:55 PM, banane wrote: This reminds me I should really go back to TDD (ducks). Ha, these errors you're having are familiar, but I can't remember the silver bullet- if there was any-that solved it. Part of me wants to say, the extension developers wrote their tests with one gem version, whereas radiant wrote theirs with another, hence the discrepancy you're seeing. This has been my biggest frustration. It's very much like the Windows DLL Hell phenomenon. I'm really going to have to dig in to why a new version of RSpec (fer cryin out loud) gets in the way of a db migration!! I just get the feeling that if I tweak one little thing, the whole house is going to come down, something I thought testing was supposed to help us get away from. Anyway, enough with the venting. I'm in the process of creating some extensions (one of the reasons for doing this was to figure out how to get started at that -- I've done a little of it but it's more cargo cult stuff than real development and I wanted to take apart a working extension to see how it ticks). The real question is how to go about setting up a development environment that I can use for this. Right now my impression is that edge is better than 0.8.1, so I'm probably going to start moving in that direction, hoping to land on 0.9 when it's ready. But I keep running into dependencies on really old testing gems. ah, well. off to bed. Have Fun, Arlen -- In God we trust, all others must supply data ___ Radiant mailing list Post: Radiant@radiantcms.org Search: http://radiantcms.org/mailing-list/search/ List Site: http://lists.radiantcms.org/mailman/listinfo/radiant Radiant: http://radiantcms.org Extensions: http://ext.radiantcms.org
