On Fri, Apr 17, 2009 at 9:26 AM, A B <gentosa...@gmail.com> wrote: > First I wish to come clean with one thing: I'd like to see LyX being > used by the ms word people in the industry, so my take is: do not > trust files someone send to you (see earlier discussions on > collaboration) > >> 1) I like the lyx format as it is BECAUSE it is not compressed, so I >> would definitely not change the default format. > > Yes, that is very nice. Easy to generate lyx files from scripts, etc. > Very nice and worth saving.
Exactly. > >> 2) I like the idea of an export format, which effectively compresses >> all files necessary into a single file. This is very nice for >> archiving final documents (lets call it .lyxa for lyx Archive). > > lyxa sounds like a nice idea, and there seems to allready be code for this. > There are two things though. lyxa should, as I see it, keep everything > in one place, all original paths will be removed, and it will be > "everything in one directory" or possible one subdirectory for images, > one for lyx files, one for bibtex stuff, one for .... and so on, as > suggested. But basically everything in one place. Reason? If you need > lyxa format, you are either archiving it for backup or sending it to > someone. The recipient are never going to have the same directory > structure as you. That will only mean failure to try to achieve. The > easier way is to just loose the exact location if you are going for a > lyxa file. > The only time I can see that the exact location is important is when > the images are changed and you don't want to update them manually in > the lyxa file, but hey, then you are probably not receiving new lyxa > versions back in collaboration with someone, so just keep your > original structure and generate a new lyxa file if you need to send it > to someone. > I hope I've managed to make my point clear about that keeping the > exact structure in a lyxa format is futile. > > >> 5) .lyxa should contain information of the original location of the >> foles on the system where it was created, to be able to update the >> files not in the subdirectory. > I say no. You do not want to trust someone on this, either sending > your paths or receiving paths from someone else. If you trust your > coworkers, set up subversion, use the same account, make it world > readable... etc. > > >> 6) One should be able to open a .lyxa file (which would modify the >> files in the .lyxa but not the original location information (from 5)) >> or imported (showing the differences of the files in the .lyxa and the >> original files and update the files when confirmed from the ones in >> the .lyxa) >> >> In this way, the .lyxa could be used as a colaborative tool (original >> author exports, sends .lyxa to other authors, they open it, save it, >> send it back, oroginal author imports it and confirms the files which >> should be overwritten) and as an archive tool of finalised documents. > > As I said no to replacing existing files with files from a lyxa file, > I say no to that, but I say yes to colaboration. > I don't want to start another flamewar about collaboration, but from > my viewpoint, don't trust your enemies lyxa files, add a "diff this > new lyxa file against this lyxa file that I have currently open and > display differences so I can accept or reject them"-button :-) > This will work also when you are working with total trust. I probably did not make myself clear: I definitely do NOT want that it is automatically replacing the original files, only AFTER confirmation, which could be done after showing the differences. > > If someone includes an eps file that generates fractals and overload > your cpu, well that is kind of hard to detect and I guess you'll have > to live with it. > > IMHO: > The lyxa format is what Lyx misses, if you have to send more than one agreed. > file to send text and image, you are alienating 99% of the population. > Then comes the question... should .lyx be available as a file format > or an internal hidden format and everything is saved in lyxa format? > But that can probably be decided in the future. > Just one idea to make the conversion easier: what about offering the option of creating the file structure automatically, when a graph or bibtex is inserted? Under Linux, this could be done via links, under Windows probably as well (.lnk files?)? i.e. when I include a file outside the ./graphs directory, a link to the original location is created? Rainer -- Rainer M. Krug, Centre of Excellence for Invasion Biology, Stellenbosch University, South Africa
