Comments on: An Exposition on Specific Time Saving Code http://blog.afoolishmanifesto.com/archives/1282 fREWdiculous! Mon, 16 Jan 2012 12:16:26 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: fREW Schmidt http://blog.afoolishmanifesto.com/archives/1282/comment-page-1#comment-974 fREW Schmidt Fri, 29 Jan 2010 18:34:25 +0000 http://blog.afoolishmanifesto.com/?p=1282#comment-974 @nperez: the main reason we haven't gone that route is that our grids are certainly not homogeneous. I briefly considered doing what you are talking about, but I just am not convinced that we would have the flexibility that we need. The same goes for DBIC::API. @nperez: the main reason we haven’t gone that route is that our grids are certainly not homogeneous. I briefly considered doing what you are talking about, but I just am not convinced that we would have the flexibility that we need. The same goes for DBIC::API.

]]> By: nperez http://blog.afoolishmanifesto.com/archives/1282/comment-page-1#comment-973 nperez Fri, 29 Jan 2010 16:47:05 +0000 http://blog.afoolishmanifesto.com/?p=1282#comment-973 That's interesting that you do a lot of the code gen inside Perl. For our platform, we use TT instead fo spit out the the record format, columns definitions, etc for the data store and grid. Mostly, I end up passing a result source directly via the stash to the template, and it is only a simple matter at that point of iterating the columns. DBIx::Class has an underdocumented "feature" that you can put pretty much whatever you want into the add_column statement, so I attach a small "hint" hash that contains things like whether it should be displayed in the grid, its display order, etc. This lets me very easily "bleed" implementations from the DB up to the display layer. In the end, our datastores and grids are restful and we have them talking to DBIC::API. All of the setup to get to that point is just an index method to spit out the initial page. I had toyed with at one point going even further and disconnecting the UI completely from TT and have it query the backend for its config (you can attach metaData to your results and it will reconfigure the record fields automagically), but time constraints left me with the TT approach to convey the structure. That’s interesting that you do a lot of the code gen inside Perl. For our platform, we use TT instead fo spit out the the record format, columns definitions, etc for the data store and grid. Mostly, I end up passing a result source directly via the stash to the template, and it is only a simple matter at that point of iterating the columns.

DBIx::Class has an underdocumented “feature” that you can put pretty much whatever you want into the add_column statement, so I attach a small “hint” hash that contains things like whether it should be displayed in the grid, its display order, etc. This lets me very easily “bleed” implementations from the DB up to the display layer.

In the end, our datastores and grids are restful and we have them talking to DBIC::API. All of the setup to get to that point is just an index method to spit out the initial page.

I had toyed with at one point going even further and disconnecting the UI completely from TT and have it query the backend for its config (you can attach metaData to your results and it will reconfigure the record fields automagically), but time constraints left me with the TT approach to convey the structure.

]]>