We never said it wasn't possible to put C++ code in the kernel, however there are many issues that has to be taken care of. One can read: http://be.befaqs.com/aboutbe/benewsletter/volume_III/Issue51.html#Insight (Hey, thanks to BeGirl who corrected the links in this mirror; it was a case issue) En réponse à philippe.houdoin@xxxxxxx: > Bruno, then Axel wrote: > >> Also it hasn't been proved that BFS can't be dinamically resized. > >> At least it is possible to do that if it would not mean changing > >> the block size. It is only that no one coded a tool to do that > >> yet. > > > > Yep, I considered it once, and wouldn't I have to replicate the > > real thing, maybe I would already have done it. > > The only trouble-maker with resizing a partition is the block > > bitmap, which will grow, so that everything behind it (the log > > starts immediately after the bitmap, then come the indices) have > > to move until they find a place. > > Another great tool would be a cdrecord's makeisofs-like tool for > OpenBFS. > I mean, a tool able to generate on-the-fly a all-packed BFS image, > from one or many existing directories & files. > > Whe then could burn BFS images on cd/dvd-rom "on the fly", which would > > be really great... > > I guess now that the source for [Open]BFS fs are available to everyone, > > someone will step in to do this tool... :-) > > On using Parted, keep in mind that's a GNU open source software, and > under GPL license. If someone want to add dynamic OpenBFS resizing > support, > it may generate a license issue here. > [Grrr, did I hate these licenses issues everywhere...] > > Off-topic, I don't know we could use C++ classes in R5 kernel-land. > Did it works only because you don't have to link the C++ code against > any shared library? > > -Philippe > >