Roland Levillain wrote:

Le 17 mars 08 à 19:39, Ugo Jardonnet a écrit :

...

https://svn.lrde.epita.fr/svn/oln/trunk/milena

Index: ChangeLog from Ugo Jardonnet <ugo.jardonnet@lrde.epita.fr>

Icp legacy. algebra:: now contains types vec, mat and quat.

* tests/metal_mat.cc: . * tests/metal_pow.cc: . * tests/all.cc: . * tests/mat.cc: . * tests/metal_vec.cc: . * tests/core/h_vec.cc: . * tests/core/tr_image.cc: . * tests/stack.cc: . * tests/dpoint1d.cc: . * tests/dpoint2d.cc: . * tests/dpoint3d.cc: .

[...]

First of all, please pay attention when filling ChangeLogs: all these entries deserve a proper comment!

Then, I don't see why everything from metal:: has to go to algebra::, in particular metal::pow<>, which actually is a metaprogramming facility!

Last, if you move something to algebra::, then the path of the file should reflect this change too, as well as the header guards.

I don't have time to discuss this much tonight, but we must definitely fix this! (You broke some tests I was working on!).

I'm bitterly sorry, :S The script I used. ---- script.sh : (dans milena) #!/bin/sh rm -f tmpffff for i in `find . -name '*.hh'` do sed -f sed.cmds < $i > tmpffff cat tmpffff > $i rm tmpffff done ---- ---- sed.cmds : s/metal::vec/algebra::vec/g s/metal\/vec/algebra\/vec/g s/metal::mat/algebra::mat/g s/metal\/mat/algebra\/mat/g ---- I may revert the patch tomorrow... For the Changelog I don't understand no more. sorry again.