On Wed, Feb 05, 2003 at 01:49:31PM +0100, Jean-Marc Lasgouttes wrote: > Andre> It is too late during any freeze. With your proposal of "stuff > Andre> to be fixed under all circumstances" we just run into a > Andre> deadlock: Some problem will be discovered in a freeze, nothing > Andre> clean can be done about it because of the freeze, so it won't > Andre> be fixed, so we can't ship, so we stay in freeze, we lose > Andre> developpers and potential contributors (I think I have seen > Andre> three or four sensible patches from "new" people which have not > Andre> been applied due to the freeze. That's "encouraging"...) > > This may be true about big bugs in the kernel. But don't forget that > we have manu small bugs/regressions that have been introduced in 1.2.0 > and that never got fixed because we have more ambitious things to do. > This `small cleanups' phase is also very necessary to have a polished > release. There has been important work done during these last months.
I do no deny the necessity of "small" fixes like the \xleftarrow fix I just commited. This was a regression simply by oversight and was fixed within three minutes. However, there are a few area that "just are not right", and some things that look like "small cleanups" are in fact just a method to hide real problems. Some of them just won't be needed if the kernel were right. The "can't scroll/jump over large insets" thing. This is a nesting problem. The undo problem is most certainly a problem of not having proper copy-semantics somewhere in the kernel. This is a probably a structural problem of the paragraph list. Math <-> Rest interaction is strange (from a user's perspective: Stuff can be copy between a table cell and a minipage but not from a footnote into a math inset). This is a structureal problem. There are lots of them. Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
