Hi Dirk! On Mon, Dec 04, 2006 at 02:40:57PM +0100, Dirk Nehring wrote: > > 2. Why revert r976? > > 3. Those collection specific defaults are part of the new style for > > defining Package Collections, i.e. on a "free willingly" and "default > > if not set" base. So if you really want to revert it, you also have to > > put the selects back into Config.in.collections. > > But this doesn't work with existing .config files or if you change your > preselection after saving. For this, forcing some core packages are > better in my eyes.
Well, in my idea of Package Collections they are useful for people setting up everything for the first time and not being absolutely sure what to choose. The only alternative is forcing a selection of those packages contained in a collection, so you will be actually forced to use dnsmasq (or at least have it installed, irrespective of it being used or not). I actually did change the way Package Collections work because wbx and me agreed on it being the better one. As we all surely know, more freedom leads to greater power, and with great power comes great responsibility. In other words, though users can more easily shoot holes into their feet, they are free now. Free as in "FreeWRT". ;) Recapitulating I would suggest to discuss this further, as it's absolutely not clear which way to go. Especially as we are still searching for alternative methods of getting a set of packages selected easily (maybe also without `conf'). > Config.in.collections must be fixed therefor (by me). What was wrong? I expected it to be working (as long as one rm's his .config before switching to/between collections). > > 4. Indeed, I did intentionally remove those "default m if devel" > > statements, on wbx's behalf. So better consult him for that (if you > > didn't already do so). > > Sorry, was in the old patch. Sorry, didn't get this one. What was in the old patch? Which old patch? Waah, sounds like "train station". > > 5. This fits perfectly into the "update-patches discussion", as everyone > > fears too less documentation and therefore unmaintainable code. Not > > always, but in this special case four words are a little less docs, > > don't you agree? ;) > > 100% agreement. svn is not a replacement for patch documentation, you > will loose all information after making dist files. Most developers are > lazy document writer, so an enforcement (patch naming) is a must have. Hmm. I think this time you didn't get it. ;) I was actually speaking of the svn log, as for everyday's changes there is just no other way to ship some documentation than providing verbose commit messages. So we could have saved at least two emails if your log wouldn't have been as minimalistic as that. (Barring the other questions I anyway have.) Also this is not about a specific package/patch, but about changes to buildroot itself (which in my eyes is the worst documented part of all ;). Greetings, Phil
pgpoMrtM7X6Si.pgp
Description: PGP signature
_______________________________________________ freewrt-developers mailing list [email protected] https://www.freewrt.org/lists/listinfo/freewrt-developers
