KDirModelV2, KDirListerV2 and UDSEntryV2 suggestions

Frank Reininghaus frank78ac at googlemail.com
Tue Feb 5 16:54:25 UTC 2013


Hi,

2013/2/5 David Faure:
[...]
> Another possible conclusion: who in their right mind puts 500.000 files in a
> directory? :-)

I'm glad to see that I'm finally not the only one who has such thoughts ;-)

I might be repeating myself, but even though ensuring that everything
works efficiently with many files is certainly a valid goal, I don't
think that the "half a million files" use case should dictate the
design of the involved classes. And no, I don't believe that
optimising for that case will be beneficial for every other use case
as well. Just one example: if you want to minimise memory usage and
thus lazy-load everything, you have to sacrifice the O(1) lookup for
URLs, which is provided by a hash. Sure, one can in principle live
without that, but sometimes, such a cheap and simple operation comes
in handy. If it's dropped, quite a bit of code has to be changed and
replaced with more complicated and slower alternatives. And of course,
turning well-tested code upside down will require a lot of work and
cause new bugs.

Best regards,
Frank


More information about the Kde-frameworks-devel mailing list