# |
Feb 19th 2009, 14:55 |
alkemann |
this is way 45% of support questions are about acl? :p |
# |
Feb 19th 2009, 14:55 |
jperras |
I'm not pro acl at all. Often times I think it's the wrong tool for the job, and will not hesitate to say so |
# |
Feb 19th 2009, 14:55 |
alkemann |
a user_profile that is bound to bakery sounds like an ok solution |
# |
Feb 19th 2009, 14:55 |
AD7six |
acl isn't hard to use though, it's actually very easy |
# |
Feb 19th 2009, 14:54 |
AD7six |
ADmad: or maybe role would be the same class as app_profile - where you'd also store "email me replies to my comments" and other settings like that. |
# |
Feb 19th 2009, 14:54 |
alkemann |
jperras: sounds like you are very pro the ACL, so i dont know what to say other that this will be much more effective and easier to work with /shurg |
# |
Feb 19th 2009, 14:53 |
AD7six |
ADmad: a role table in the bakery db. user_id, role (group, whatever it should be called) to be used as the lynchpin for acl or the group-finding-model otherwise |
# |
Feb 19th 2009, 14:52 |
alkemann |
gwoo wants this app to built as a standalone and not especially made for cakephp.org though |
# |
Feb 19th 2009, 14:52 |
jperras |
here's what I don't get alkemann: you and ADmad have spec'ed out a permissions system which is nearly identical to ACL in 'controller' authentication mode using a static config.ini file (instead of db), and you want the bakery to roll their own perms. system instead. |
# |
Feb 19th 2009, 14:52 |
AD7six |
so user | app data needs to be seperated, and the user-group link would be on the app data side of the divide |
# |
Feb 19th 2009, 14:52 |
ADmad |
AD7six: yes this was jotted up b4 we had that directive... what changes/alternative do you suggest |
# |
Feb 19th 2009, 14:51 |
alkemann |
yes, i agree that this feature would be good. |
# |
Feb 19th 2009, 14:51 |
AD7six |
whoo'ps. |
# |
Feb 19th 2009, 14:51 |
AD7six |
and any other apps built going forwards |
# |
Feb 19th 2009, 14:51 |
AD7six |
and any other apps that are built going forwards |
# |
Feb 19th 2009, 14:50 |
AD7six |
the same user table is used in the book - the book's permissions are seperate and different |
# |
Feb 19th 2009, 14:50 |
AD7six |
alkemann: you weren't here when I came in: something that I consider important is that the user data should be seperate from the apps data and logic |
# |
Feb 19th 2009, 14:50 |
jperras |
AD7six: if Group hasMany User, where else would it be? |
# |
Feb 19th 2009, 14:49 |
alkemann |
yes. important part of the solution |
# |
Feb 19th 2009, 14:49 |
AD7six |
for one the group_id field is in the users table |
# |
Feb 19th 2009, 14:49 |
AD7six |
oo don't like that so much ;) |
# |
Feb 19th 2009, 14:48 |
jperras |
bottom of the wiki page, AD7six |
# |
Feb 19th 2009, 14:48 |
alkemann |
http://thechaw.com/bakery/wiki/spec/users/Group_permissions |
# |
Feb 19th 2009, 14:48 |
AD7six |
I see a description of permissions |
# |
Feb 19th 2009, 14:47 |
AD7six |
I don't see where it says how you're going to store and query permissions? |
# |
Feb 19th 2009, 14:47 |
alkemann |
the one* |
# |
Feb 19th 2009, 14:47 |
alkemann |
AD7six: the we have specced already |
# |
Feb 19th 2009, 14:46 |
AD7six |
alkemann: what solution would you suggest |
# |
Feb 19th 2009, 14:46 |
jperras |
which the acl behavior supports (ini configuration instead of db config) |
# |
Feb 19th 2009, 14:46 |
jperras |
with a static config file |
# |
Feb 19th 2009, 14:46 |
jperras |
you're basically implementing a flat tree acl in that description |
# |
Feb 19th 2009, 14:45 |
alkemann |
we dont need it. |
# |
Feb 19th 2009, 14:45 |
jperras |
I don't see a reason to not use acl here |
# |
Feb 19th 2009, 14:44 |
alkemann |
"point to a random" |
# |
Feb 19th 2009, 14:44 |
alkemann |
as far as i am concerned, if you can find a reason to not use acl, take it. and the since we dont need to be able to point to give a random user a random right to a random asset, acl is overkill |
# |
Feb 19th 2009, 14:44 |
jperras |
of course not. but it's already written, heavily tested, and included in the cake core |
# |
Feb 19th 2009, 14:43 |
alkemann |
so acl is the only valid implementation of permissions? |
# |
Feb 19th 2009, 14:43 |
jperras |
well, if it sounds like a duck, quacks like a duck, why not use a duck? |
# |
Feb 19th 2009, 14:42 |
alkemann |
yes. permission implementations are bound to share descriptions :p |
# |
Feb 19th 2009, 14:41 |
jperras |
control user permissions based on their group, and the controller action that they are attempting to use |
# |
Feb 19th 2009, 14:41 |
jperras |
alkemann: that first paragraph sounds exactly like what acl does |