Shawn Walker wrote: > 2008/7/10 Brock Pytlik <[EMAIL PROTECTED]>: > >> Danek Duvall wrote: >> >>>> image.py >>>> 132-133 Could you elaborate on the difference between "required_subdirs" >>>> and "image_subdirs"? (and if it's what I believe it is, can you place >>>> "index" into the image_subdirs instead of the required subdirs?) >>>> >>>> >>> Being in required_subdirs means that it needs to be present in order for an >>> image to be identifiable. Being in image_subdirs means that it will be >>> created when the image is created. That's all in the comment above. >>> You're saying that index should no longer be a required directory in an >>> image? >>> >>> >> That's correct. If it's not there, it can (and will) be regenerated on >> the fly unless the next action is a pkg search, in which case the user >> will be directed to rebuild the index. (This could be done >> automatically, but it seemed better to have the user make this choice >> since indexing is (at least) an order of magnitude slower than searching.) >> > > I'm not terribly keen on the idea that the user needs to even know > that an index exists. > > That seems like something that should be invisible to the user. > > With that said, I think it would be perfectly reasonable to let the > user know that we're doing extra processing and that the operation may > take longer the first time. > > In general I agree. Having them run rebuild-index is only necessary in the event "something bad happens." I continue to believe that asking them to run this command rather than having search (especially their first search under the new system) suddenly take 40s instead of 1s satisfies the principle of least surprise. If it's generally preferred, I can print a warning message, reindex, and then run the search again. Either way, it shouldn't be part of the required_dirs.
Brock _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
