Comments for Code Colossus

Comment on Using ActiveModel::Name to simplify URL generation by benhughes

benhughes — Wed, 28 Dec 2011 17:54:32 +0000

I dug into the Rails source to find out what exactly was generating those polymorphic_path URLs based on AR models, and saw it was ActiveModel::Naming’s model_name, not class.name.

Comment on Using ActiveModel::Name to simplify URL generation by andy

andy — Wed, 28 Dec 2011 16:35:10 +0000

thanks so much. been finding this solution for 2 days. btw how do you find out “self.model_name”?

Comment on ActiveRecord Drafts with has_draft by benhughes

benhughes — Sun, 15 May 2011 17:08:58 +0000

Thanks, fixed!

Comment on ActiveRecord Drafts with has_draft by Ramon

Ramon — Sun, 15 May 2011 13:04:33 +0000

Looks like your link has moved to https://github.com/rubiety/has_draft

Comment on Using ActiveModel::Name to simplify URL generation by jaye

jaye — Mon, 02 May 2011 09:35:21 +0000

beautiful. Thanks

Comment on message_block: a error_messages_for replacement for flash message and model error handling by Bill

Bill — Thu, 10 Feb 2011 04:45:18 +0000

It’s here:

https://github.com/rubiety/message_block

Comment on Why you should always validate maximum lengths in models, and how to do so easily by Eric Anderson

Eric Anderson — Sun, 02 Jan 2011 20:55:21 +0000

Doing validations based on the database schema is something I have been doing for a long time in Rails and can’t imagine doing it differently. Just way to many validation statements to write otherwise.

I used to use schema_validations plugin[1] but the developers stopped supporting it a while back. Plus it was very heavy-weight. Recently I have been using enforce_schmea_rules[2] which I even contributed patches too (to make it Rails 3 compat). I think it needs better testing still and the code is a bit old but it works well and might be something to look into to get ideas.

1. https://github.com/harukizaemon/redhillonrails/tree/master/schema_validations
2. https://github.com/twinge/enforce_schema_rules

Comment on Why you should always validate maximum lengths in models, and how to do so easily by benhughes

benhughes — Sun, 02 Jan 2011 19:21:39 +0000

Ben – Great points about the numeric columns. I haven’t even considered them being a problem as well, but indeed they are. Thanks for your commit adding this to the project; I’ll have to add supporting tests and modify the README to reflect the changes, then I’ll push them up and release in a few days.

Comment on Why you should always validate maximum lengths in models, and how to do so easily by Ben Simpson

Ben Simpson — Sun, 02 Jan 2011 18:47:04 +0000

I agree that point #2 is a big issue (especially with MySQL). Since the truncation is silent, you wouldn’t know its gone until you query again. No exceptions are thrown, and ActiveRecord happily goes on about its day.

As an interesting note, Sqlite supports dynamic typing, which I discovered makes it very hard to get it to truncate a value since it does its best to accommodate (From their website: “You can put arbitrary length strings into integer columns, floating point numbers in boolean columns, or dates in character columns”.) While that takes care of point #2, it compounds the Denial Of Service and data integrity issues.

Your plugin is a great way to ensure that text and string datatypes are not truncated, you are helping prevent a DOS attack, however the mentioned issues seems to occur with numerical datatypes on MySQL as well.

I forked your project on Github to play around with, and pulling back the curtains on the migration transitions to database column types is horrifying. If precision and scale aren’t specified in a decimal column definition for a migration, Rails treats it as a type integer. Floats, decimals, and integers all silently truncate. Floats truncating are probably fine, since this is the design idea behind a float rounding.

Decimals should have precision and scale enforced before save. Integers should also report if there is possible truncation. Hopefully the code I pushed is helpful in some way.

Great work!

Comment on Why you should always validate maximum lengths in models, and how to do so easily by Michael Raidel

Michael Raidel — Sun, 02 Jan 2011 14:58:51 +0000

Very good idea Ben!