On Friday 31 August 2007 15:55:26 Richard Heck wrote: > Bo Peng wrote: > >> 1/ if it is not done already, support for reading/writing compressed > >> lyx files should be removed: it has no interested wrt .zip file > >> version. > > > > I agree. It does not make sense to keep two compressed version. I am > > not sure if lyx2lyx decompress compressed lyx file though. > > Jose answered this one for me: > def choose_io(self, input, output): > """Choose input and output streams, dealing transparently with > compressed files.""" > Presumably this function could be modified to deal with the new format.
That is the plan. It implies very few code actually. :-) > >> 2/ is the new zip base version designed such that a "file" utility can > >> guess its type based on the first few characters of the binary file? > >> It could be doable if the first file (the manifest?) is stored instead > >> of deflated. > > > > Yes. It is a regular zip file. It is not easy to get .lyx or manifest > > version without unzipping it first though. > > But file will return that it's a gzip file. I've tried it. I get a different result: bug-1.lyx: \012- LaTeX document text bug-2.lyx: Zip archive data, at least v2.0 to extract bug-3.lyx: gzip compressed data, was "bug-3.lyx", from Unix, last modified: Fri Aug 31 00:28:18 2007 Notice that 2 and 3 show different results. > >> 4/ I really thing that we should change the extension for bundles (I > >> like .lyz personally). > >> > >> a) it allows for automatic processing of files > >> b) when an end-user receives a .lyz file, he knows that everything > >> needed is in there. With a .lyx file it is expected that some things > >> are missing. > > > > If we remove the compression feature, this makes sense. Still, what > > about my 'open file.lyx, enable embedding, save to file.lyz' scenario? > > This looks weird to me. > > I can see wanting embedding to be more transparent, but it's a > trade-off. And maybe it's worth seeing right away that it's an embedded > file, anyway. > > By the way, let it not be lost in this discussion how useful this will be. :-) > Richard -- José Abílio
