Actually I doubt thet there is a 'standard' for any filenames - they are just stored as a series of bytes about the only 'direction' is that backslashes be stored as slashes! Unicode is not as difficult as MBCS (WinXP/NT), to work with them you must know code page they were 'encoded' with! - Russell Peters ----- Original Message ----- From: "James Turner" <james.d.h.turner@xxxxxxxxxxxx> To: <delphizip@xxxxxxxxxxxxx> Sent: Friday, November 01, 2002 12:42 PM Subject: [delphizip] Re: New format > > Gerard, > > You may very well be right in your analysis that the Zip format is less than > ideal for your requirements, however, I would certainly not be confident in > actually agreeing with you unless I knew exactly what you want to do. In any > case, there are several archive formats around these days so it may be the > case that one of these would serve your requirements better. I am new to > this field and know nothing of the likes of RAR archives but if you explain > in detail exactly what you want to do, someone on this list may well be able > to point you in the right direction. > > So far as designing a new format is concerned, I would suggest you sit down > with a pencil and paper for a couple of hours or so and scribble down your > thoughts and the necesary algorythms for manipulations such as rename. I may > be totally wrong, but I suspect that there are only two significant changes > that might be useful. > 1 : Strip variable length data from the file header blocks (or eliminate > them altogether). This would permit easier renaming. > 2 : Create some sort of allocation unit system (rather like disc sectors > but variable length). This may permit faster delete and overwrite operations > but has major drawbacks with regard to overall compression efficiency. > > In other words, perhaps you are being unrealistic with regard to what you > want. However, there are certainly subscribers to this group who know vastly > more about this stuff than I do. > > Personally, other than the fact that zip headers and records use 32bit > integers instead of 64bit ones and there is no standard method of using > unicode filenames, I think the zip format is ok. However, large archives are > undoubtedly troublesome if high speed manipulations are required. > > Perhaps what you need is a virtual compressed disk system like > DriveSpace/DoubleSpace. I seem to recall that you could only allocate a > fixed amount of disk space for such virtual disks but I could be mistaken or > someone may have designed a system that eliminates this restriction. Of > course, development in this area has pretty much stopped because hard disks > have such huge capacities these days. Perhaps also, the DBX file format > (used by Outlook Express amongst other programs) might be right for you but > I have no knowledge in this area. > > Sorry if my thoughts on this subject are not very helpful. > > James Turner > > PS > My website will be going live in a day or so. After that I had planned to > start work designing my own minimalist zip implementation. However, as Eric > has just pointed out, there is a Nanozip to consider. I've only glanced at > this so far but it looks like it is well written and it may cover all my own > requirements for a pure Delphi zip solution. In which case I shan't bother > writing my own version. > > ----- Original Message ----- > From: "Gerard" <gslurink@xxxxxx> > To: <delphizip@xxxxxxxxxxxxx> > Sent: Thursday, October 31, 2002 10:32 PM > Subject: [delphizip] New format > > > > > > Hi all, > > > > I am using now delphizip for several months and I am having difficulties > > all the time achieving the things I have in mind. I mostly manage but it > > takes quite some tricks to get it done. I don't think that this is because > > of the existing code. Chris, Eric, Russell and the others did a great job. > > I think it is an essential shortcoming of the way zip files are designed > > and urge to stay compatible with it. I have the impression that the zip > > format was designed for batch file processing and not so much for single > > file operations, (ie to rename a file you must to rewrite the whole zip > file). > > > > A thought came to my mind. What about a new format, which uses still the > > proven compression methods (and allows for adding new methods) but stores > > the files in a different structure, which makes single file access more > > convinient and allows for all sorts of options which are now difficult to > > achieve. Something like a 'file-object database'. > > > > We would lose compatibility with the zip format, but for many > applications, > > except for the Winzip clones (this is not meant negative - I have my own > > Winzip clone :^), this might not be such a big problem. > > We could use very much of the experience we have to design a new format. > > > > What are your thoughts ? > > Does this sound crazy, asking for trouble, an effort only for someone who > > has nothing else to do or might this be a good idea ? > > What are the fundamental problems and shortcomings you encounter, what are > > your wishlists ? > > > > Please your reactions, > > Regards, Gerard. > > > > > > > > > > > > > > >