Scribit Steve Greenland dies 04/04/2005 hora 07:15: > > - what problems do thsi random order could weed? > Unnoted dependencies that just happen to be fulfilled due to a > consistent (though arbitrary) application order. By applying in a > different order each time, you should trigger an error fairly quickly. I don't find it very sane to be forced to deliberately trigger problems on the user's system to find bugs. Wouldn't it be more appropriate in a package testing tool? Maybe lintian, with a test of all combinations of independant patches (or a more intelligent subset) to see if something fails. When I unpack a source, I don't want it to fail to help a careless maintainer to find the flaws in one's packaging... > > - won't it be more difficult to trakcs bugs if it isn't predictable? > If you get an error during the patching process, it should be fairly > easy to determine that it's an un-marked dependency, and then find it > by hand. That's what you think. I'm not so sure. > You can also impose arbitrary dependencies among your supposedly > "independent" patches until you find the troublesome combination. Urgh. Are you doing computer science or cooking? Couldn't the application write somewhere the order of the patch that it just used?! Easily, Nowhere man -- nowhere.man@levallois.eu.org OpenPGP 0xD9D50D8A
Attachment:
signature.asc
Description: Digital signature