Bastian Friedrich wrote:

Hi,

On Thursday 04 October 2007, Bogdan-Andrei Iancu wrote:

...

db_berkeley
mysql -> db_mysql
postgres -> db_postgres
dbtext -> db_text
unixodbc -> db_unixodbc
flatstore -> db_flatstore

... perlvdb -> db_perlvdb?

Bastian

Right :)

Regards, Bogdan

On 10/04/07 13:36, Henning Westerholt wrote:

On Thursday 04 October 2007, Bogdan-Andrei Iancu wrote:

...

Hi,

during the reviewing process of the new DB module for Berkeley, I made proposal to make a naming convention for all the modules implementing the DB API.

I suggest something similar as for the modules implementing the MI interface, all having the "mi_" prefix. So, for the DB modules we should use "db_" prefix: db_berkeley mysql -> db_mysql postgres -> db_postgres dbtext -> db_text unixodbc -> db_unixodbc flatstore -> db_flatstore

Does anybody see any drawback or problem if we rename the module before 1.3?

This change is resonable for me _if_ it will be thoroughly carried out. That means in addition to the renaming of the directories its also necessary to

• update the existing module documentation (README files, devel index.html)
• fix the build process and the packaging
• update openser install documentation (INSTALL, wiki pages)
• add a note to the porting guide for 1.3

as personal preference, I would add here the option of still being able to use database name in db_url parameters -- looks more natural in config file

modparam("avpops", "db_url", "mysql://openser:openserrw@localhost/openser")

than

modparam("avpops", "db_url", "db_mysql://openser:openserrw@localhost/openser")

Should be easy to attempt to try to find the module "mysql" or "db_mysql"

Since we got in such topics, perhaps we should go further and define a naming policy for: - module name - main file of the module - exported functions by modules

For the first one, should be representative for the functionality brought, and in particular cases (db driver, mi transport, radius extensions, ...) should follow a clear pattern. If going for the above proposed system, first should be the category (same happens now with mi/presence/pua modules), so seems to be the appropriate one. In this way is very easy to spot the group of modules related to a category. Also, the name of the directory should be same as module name.

For the second, there are two main patters - main file name matches the name of the module - main file name is suffixed by "_mod" Several modules do not follow any of above.

For the third, I propose that functions will pe prefixed by several letters derivate from module name, maybe abbreviation, or full name if it is short.

I think these will bring more coherence and consistency in the code and config file. Also, will easy automatic tools for checking exported structures, or generating goodies docs out of C code.

If the time for this release is not enough, we can leave it as it is, but should be taken in consideration for new contributions.

Cheers, Daniel

Cheers,

Henning

---

Devel mailing list Devel@openser.org http://openser.org/cgi-bin/mailman/listinfo/devel