Sonata/Admin integration

[sylius]

Sonata/Admin integration

[pjedrzejewski]

I like Sonata, but it's quite a lot of code - really big deps. I never used it in big project, but I know that as the project grows, it can become a bit problematic to extend / customize. I don't think we'll ever integrate any admin generator bundle into the core. Resource bundle is much simpler and so far does the job. :) Still, I think it's super simple so put Sonata admin for existing app / custom models + Sylius. I never heard about any conflicts. Thanks for raising this topic!

[pjedrzejewski]

I leave it open if anyone want to participate / discuss. :)

[marcospassos]

I agree. Sonata is not a good option for this project, it has a lot of dependency and requires a initial knowledge. I'm totally against.

[umpirsky]

My biggest fear is that it can stand in the way. So -1
On 7 May 2013 14:35, "Marcos Passos" [email protected] wrote:

I agree. Sonata is not a good option for this project, it has a lot of
dependency and requires a initial knowledge. I'm totally against.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/30#issuecomment-17539275
.

[kingcrunch]

I respect the decision, but just the get that right:
I started this discussion, because my thought was, that you can save some effort, if you rely on existing solutuions you don't need to maintain yourself (on the one side) and also to avoid X different Admin-Panels because of NIH, but this arguments makes me wonder a little bit:

• It has dependencies and
• One must get used to it?

(Beside: One cent for every time I hear "We should build it on our own, because it's much simpler 😉)

Still, I think it's super simple so put Sonata admin for existing app / custom models + Sylius. I never heard about any conflicts. Thanks for raising this topic!

I don't know exactly, if I understand you. Maybe (if I find some time 😉) I'll look at some kind of sonata-admin-integration, but I don't think it's worth the effort...

[sstok]

I starting using the SonataAdminBundle, but soon found it was just to much, the application I'm building is very specialized and Sonata made this way to complicated. Using it for the backhand was good enough but the customer section would be hell because of the complex routes and plug-ability of things, but as Sonata provides some good things I started extracting things like the route builder and batch processing.

Now that I find this beauty, things changed for the better!! 😄 I can use this bundle for most things that Sonata is to bloated for, but use the things that Sonata did a great job at and combine them.

The goal is to combine the extracted thing to an open source bundle, but I'm not completely sure if that is going to work so don't except anything anytime soon. 😦

[sstok]

Update.

The good news is that for Datagrid handling (including batch actions) I found https://github.com/Abhoryo/APYDataGridBundle

The bad news is that things like automatic route loading, breadcrumb building, navigation and such would require more work and some sort of Admin class, doing this will bound it to much to an implementation (as I'm using the MopaBootstrap bundle) and parent handling for the route generation is not something everyone uses the same way, so instead I'm just going to work on this and make Gist of things when I'm done 😄

[norberttech]

@sstok check https://github.com/fsi-open/datagrid-bundle for data grids and https://github.com/fsi-open/datasource-bundle for data collections that are used by data grids.

[sstok]

I have actually looked at both :) from what I could see so far the fsi-open/datagrid-bundle has no support for handling batch actions (only a column type).

[norberttech]

Actually fsi-open/datagrid has no support for handling any kind of actions by default. It is like the form component, it gives you form object with methods like bind but you need to write request handling on your own.

[sstok]

Ah interesting, looking at it (allot better) everything I wand to do is perfectly possible :)

[dyjo]

-1

I have explored Sylius & Sonata integration, and found it best to keep the two entirely separate; even considering using separate kernels for the two (see: Sylius/Sylius#84).

The first problem I see with backpedaling and using Sonata for the backend is it seems like it would be a lengthy undertaking that would delay a "ready-to-go" Sylius/Sylius.

The second problem I see is that a lot of the bloat associated with Sonata doesn't make sense for Sylius (only a select few entities would benefit from datagridding, products, options, etc.)

But, what I would like to see is something like my RFC above. Sonata is great for content, so maybe using some of those components to build a paid-content version of Sylius.

[marcospassos]

-1 again