Am Sonntag, 9. Juli 2006 11:53 schrieb Lars Gullik Bjønnes: > Georg Baum <[EMAIL PROTECTED]> writes: > > | Am Sonntag, 9. Juli 2006 01:28 schrieb Lars Gullik Bjønnes: > | > The Change to insetgraphics is obvously ok (as long as we don't use > | > boost::path directly) > | > > | > The change to FileName is really ortogonal and quite a bit more > | > controversial. (Not really, but...) > | > | They are not orthogonal. The change to insetgraphics alone would have led > | to every file opened twice for checking whether it is zipped or not. > > But that is an implementation detail? Sure it is slow, but still > correct. So the addition of a zip cache is only an optimization.
I see it a bit different: The patch to insetgraphics.C alone would have introduced a slowdown, so I would not want it if the zip cache is for some reason not wanted. > They are related, but not dependant. Lars, what do you want to achieve with this discussion? The patch is small enough to understand it easily. You say the patch should have been splitted because a part of it might be controversal, but then you don't say what is wrong with it and that you think it would not be reverted? Sorry, but if you want us to split up patches only for theoretical reasons without any practical benefit then you will eventually loose all contributors (including me). I do agree with the "only one thing in one patch" approach. I know the benefits, and I try to follow it (otherwise I would not have done this particular change alone but it would have been part of a bigger patch). I even would understand your comment if you had real concerns, but since that is not the case it only steals my time. Georg
