I have an application that requires some per-tenant configuration. What I would like to do is load configuration for specific bundles using parameters embedded in the route:
Route: /{tenantId}/some/resource/390234
where I can then load a configuration specific to $tenantId.
As far as I know this is not possible, since configuration loading in symfony2 happens on compile time.
This means every time you do a
php app/console cache:clear
or cache warmup.
Why not do this via configuration entries via database? As an alternative, use a custom class to load e.g. yaml files (don't forget caching) with your own configuration.
This problem is driving me crazy... I think I've tried every conceivable combination of Sass file, ERB file, asset helper, image helper, etc. etc. Someone please give me new ideas!
Context:
Rails apps require use of asset helpers so that when the assets are precompiled, the source will be a fingerprinted asset file. I.e., if you just called img src="X.jpg", the site in production would look for X.jpg, but the file in public/assets has actually been fingerprinted as X-as;diofua;wemfiwaejfoiawefo.jpg. The only way to get to that fingerprinted file is to use an asset helper, e.g., image_url ('X.jpg').
Right now in my live site, I'm using an asset helper, but for whatever reason, it's not pointing at the fingerprinted asset. Note that the assets are found in development (but again, that's because there's no fingerprint added in development).
Code
Image titled "classic-map.png", located in app/assets/images/galleria
Image is called from a css.erb file required in the application.css file. In the css.erb file, I have the following code:
background-image: url(<%= asset_path 'galleria/classic-map.png' %>);
For reference, http://guides.rubyonrails.org/asset_pipeline.html
Note that I'm choosing to write this as a css.erb file, hence the use of asset_path vs. asset-path. Also, I initially thought that the issue might have been in interpolation, but in the page source, the url is definitely working, it's just that it's pointing at url(galleria/classic-map.png) instead of url(galleria/classic-map-apsoidufalskjf;kasj.png)
A million kudos to whoever can help!
For what it's worth, this happened to be AGAIN, and this time I could not use the hack because I desperately needed the fingerprint. So somehow, magically, I ran a rake assets:clobber and heroku run rake assets:clobber to clean all assets, and then a straight up git push to force Heroku to do the precompilation for me. That did it, and everything works.
Now, when this happens, I clobber assets locally & in production and push, forcing Heroku to precompile remotely. Similar to #user2880239's answer. I have stopped precompiling locally and checking into git.
I sat with a Sr Rails developer who still couldn't help me fix this. But the workaround we ended up using was that we just manually removed the asset fingerprint in the public folder (since the fingerpoint is what the asset helper is meant to point to).
I.e., the file galleria/classic-map-587854758918434124.png we just manually changed back to galleria/classic-map.png and it works fine.
Note that if you do this 'hack', the next time you precompile assets, Rails will create another fingerprinted asset, so you'll have duplication unless you want to keep deleting the additional fingerprinted asset each time. For me, I don't care about the duplication, I care about not thinking about this anymore.
Did you check RAILS_ENV ?
bundle exec rake assets:precompile RAILS_ENV=production
I had the same problem you did. This blog post helped me.
What I did was change a few things in my config/environments/production.rb file, namely...
config.serve_static_assets = true
config.action_dispatch.x_sendfile_header = ‘X-Accel-Redirect’
config.assets.compile = true
Note that you might not need to 'add' any of those properties since they may be pre-set to false or merely commented out.
Then I did the heroku dance:
rake assets:precompile
git add .
git commit -m "Fix static assets"
git push
git push heroku
I'm having the same issue. I even tried the helper from the rails console from heroku, and the helper works fine there!!
$ heroku run rails console
Running `rails console` attached to terminal... up, run.8071
Loading production environment (Rails 4.1.7)
irb(main):001:0> puts helper.image_path("bg.jpg")
/assets/bg-00acfb7dbe138102509d82ac2313c24d.jpg
My final "solution" was to update config.assets.compile = true in config/environments/production.rb to fallback to the non fingerprinted image.
Hope this solution could help to someone. And if you had any real solution, please make me know!
The answer to this for Heroku is in their Pipeline docs here.
By doing clobber you are basically cache-busting all your assets and forcing all clients to reload all static assets (even if they have not changed) every time you deploy your code. That is not advisable as it means every time you deploy ALL clients will experience slow loading times until all assets get cached again.
Your css file has a dependency to your image file, so you need to tell the assets pipeline about this by putting this at the top of your css:
//= depend_on_asset "galleria/classic-map.png"
This tells sprockets that if class-map.png gets a new fingerprint then the css must also get a new fingerprint. So it will only recompile the files (and dependencies that changed).
Also for others landing here, be aware that if you are using asset_path from ANYWHERE other than a view (eg in a model) you need to prepend the full context:
ActionController::Base.helpers.asset_path('your-image.png')
More info here.
I have had livereload working with guard in my symfony site. But my assets (css/js) were in the 'web' directory. But then I learned that the best practice is to keep everything in 'src' so that you can keep all your files as a bundle. So I moved it to the Resources/public directory in my bundle under my 'src' dir and ran:
php app/console assets:install target [--symlink]
This created the necessary symbolic links from 'web' to my 'src' assets, everything seems to work if I run guard in 'src/.../Resources/public', sass, compass, however live reload no longer works. I'm wondering if it has anything to do with the symbolic links (it was working before, I'm using livereload.js) I've read here:
symfony2 assetic dump stopped watch
and here:
No route found for "GET /img/glyphicons-halflings.png"
that people have got livereload to work with symfony/guard/sass. The trick, I assume is to write the Guardfile such that it runs the assets: install or assetic:dump, which I guess is what makes livereload work? But I can't find anyone explaining how they got this set (symfony2, livereload, sass, guard) to work (maybe just the Guardfile?).
There is a bundle for live reloading assets: http://bundles.kunstmaan.be/news/automatically-refresh-your-browser-on-changes
The bundle may be the better method. But all I ended up doing was downloading liveload.js from their site, adding it a javascript include, added this to my Guardfile:
guard 'livereload' do
watch(%r{.+\.(css|html|js)$})
end
and ran guard manually.
Is there a way I can extend Assetic so it will allow me to have it load my images from a CDN which I configure somewhere else. If I can write a plugin/extension/event listener/etc which allows fine-grain control, that would be awesome.
Likewise, are there any hooks for the assetic:dump console command which I could use so when I perform a dump I can automatically push images from the local box to the CDN (since I want the images to load locally on dev)?
Basically, I want to be able to have an images sitting somewhere in on my local machine, say:
web/images/image.png
On dev, when I do something like:
asset('images/image.png')
It would spit out a URL like this on dev:
http://1.2.3.4/images/image.png
And on production it'd spit out something like:
http://my.cdn.net/images/image.png
And when I run assetic:dump, the image is automatically uploaded.
Thanks.
Check out this link: http://symfony.com/doc/master/reference/configuration/framework.html#assets-base-urls
You can add a base url in your config.yml like:
framework:
templating:
assets_base_urls:
http: my.cdn.net
Since Symfony 2.7, configuration has changed:
framework:
assets:
base_urls:
- 'http://cdn.example.com/'
check out :
http://symfony.com/blog/new-in-symfony-2-7-the-new-asset-component
I designing an programming a website using Symfony2. I completed some parts and checked the client side in both dev and prod environment. In the first time everything was fine and I saw same result in both environments, But after sum changes in views (I'm using Twig with Three-level Inheritance) such as adding sum other stylesheets in "stylesheets" block, dev environment works fine but in prod environment no stylesheets added and its using old data. I cleared cache and log manually from both path "app/cache/dev" and "app/cache/prod" ... but nothing changed. I removed all stylesheets in block and cleared chache again old data is still there!
Any suggestion?
Apart from clearing the cache, make sure you publish your assets:
app/console assets:install --symlink web
app/console assetic:dump
The problem was about removing cache files. I had been removing cache files via cPanel filemanager. It gave "all files removed successfully" but they didn't. So I found another way to clear cache. I used shell_exec:
shell_exec('rm -r app/cache/prod');