[delphizip] Re: Span volume problems
- From: "Roger Aelbrecht" <Roger.Aelbrecht@xxxxxxxxxx>
- To: <delphizip@xxxxxxxxxxxxx>
- Date: Fri, 6 Dec 2002 18:45:09 +0100
Hi Russel
For Reading
As it is implemented now in ZipBuilder
Start with ZipFileName = filename.zip
If present test if spanned, spanned means on removable disk
with volume name set to PKBACK ###
If not present start search for filename###.zip
If on hard drive search for the last volume this should be the last part
If on removable disk ask for last disk each time adapting filename
For Writing I added a bool property (can be replaced by member of set)
NoVolumeChange in that case we use the filename###.zip for each part
no difference between hard disk and removable
Another bool property WinZipNames if true
changes both Reading and Writing by using the filenumber scheme
to Fielname.z## except for the last part
I defending these schemes for two reasons
I already made it ;-)
The first part was already there but only for hard disk, so changing number
schemes will give backward compatibility issues
I use separate Boolean properties but these van very easily changed into
members of the AddOptions set
But had problems with having an AddOption influencing the Reading part
therefore choose to use separate property
Roger Aelbrecht
http://web.wanadoo.be/driehoeksw
----- Original Message -----
From: "Russell Peters" <russellpeters@xxxxxxxxxxx>
To: <delphizip@xxxxxxxxxxxxx>
Sent: Thursday, December 05, 2002 9:56 PM
Subject: [delphizip] Span volume problems
>
> I am trying to get around volume name problems but it is rather involved
> (you have to be able read them back readily)
> I thought to check in following sequence
> if not filename###.zip (## is number) assume must be last disk (error
unless
> Volume name is normal span volume name).
> if filename###.zip must be part of span set (CheckForLastDisk(EOC,false)
> returns false)
> if filename.z## must be part of span set (CheckForLastDisk(EOC,false)
> returns false).
>
> Then use new add options AddSpanNoVolumeNames and AddSpanCompat to control
> generating the zipfiles. AddSpanNoVolumeNames would make it generate
> numbered files instead of numbered volume names and AddSpanCompat would
make
> it generate WinZip type names.
> (Personally I would prefer filename.zip.nnn )
> Thinking of template using brackets thus - filename[,nnn].zip or
> filename.z[ip,nn] or filename.zip[,.nnn] where [ final , parts ].
> Russell Peters
> http://www.users.bigpond.com/russellpeters/delphizip.html
>
> ----- Original Message -----
> From: "Alistair George" <bigal@xxxxxxxxxx>
> To: "Roger Aelbrecht" <delphizip@xxxxxxxxxxxxx>
> Sent: Friday, December 06, 2002 7:18 AM
> Subject: [delphizip] Previous spanned reported spanned error
>
>
> >
> > Hello Roger,
> >
> > Just figured out what is wrong. I made the spanned set on hard disk,
then
> copied
> > the spanned volumes to CDR (this means the Volume name is wrong, or non
> existent :-(
> > What I have been doing is to write volume sequence number on the disks.
Is
> there
> > an easy way around this (can you suggest what is needed to make
zipmaster
> accept
> > unnamed volumes) as it is preferable to write CDROM directly off the
file
> > on Hard Disk for obvious reasons.
> > Cheers,
> > Alistair+
> >
> > Friday, December 6, 2002, 6:48:14 AM, you wrote:
> >
> > RA> ----- Original Message -----
> > RA> From: "Alistair George" <bigal@xxxxxxxxxx>
> > RA> To: "Roger Aelbrecht" <delphizip@xxxxxxxxxxxxx>
> > RA> Sent: Wednesday, December 04, 2002 7:33 PM
> > RA> Subject: [delphizip] Re: Maxdisksize FreeSize
> >
> >
> > >>
> > >> Hello Roger,
> > >>
> > >> >> A major weakness of the current ZIP format is its slowness in
> returning
> > RA> a
> > >> RA> file
> > >> >> within a spanned archive. Usually I make CDRW archives in
segments,
> so
> > >> RA> that retrieval
> > >> >> is within the same disk.
> > >>
> > >> RA> To be honest I think this was the way it was originlly inteded to
> be.
> > >> RA> The whole issue of wrting to CD and CDRW came up long after the
> PKZIP
> > >> RA> format was defined.
> > >>
> > >> RA> Personnly I still don't understand why the Central Directory has
to
> be
> > RA> at
> > >> RA> the end
> > >> RA> (or last disk) in fornt or onthe first disk would make things
much
> > RA> easier.
> > >> RA> But then again we are restarting the discussion of a new
format.;-)
> > >>
> > >> Would having a central directory on each disk stuff up multi-disk
> backward
> > >> compatibility?
> >
> > RA> I cannot answer that without going the procedures and maybe do some
> testing
> > RA> But as it is written today, the central directory needs to be one
and
> > RA> complete
> > RA> So putting it on each disk will not always guarantee that you have
the
> right
> > RA> disk
> > RA> in the reader, and then we end with the same problem
> >
> >
> > RA> This is a major problem with disk spanning if you have several
> > >> CDRW with data it takes forever to restore a single file but if you
> have
> > RA> each
> > >> disk 'just filled' and do several separate archives the restore
> function
> > RA> is very
> > >> fast.
> > >>
> >
> >
> >
> > >> Anyway, having the ability to write fixed size to various disks if
you
> > RA> could
> > >> impliment it without too much ado would be appreciated by me - if
> anyone
> > RA> is
> > >> against it please put your pennyworth in thanks.
> > >> Alistair+
> > >>
> >
> >
> > RA> Roger Aelbrecht
> > RA> http://web.wanadoo.be/driehoeksw
> >
> >
> >
> >
> >
> > --
> > Regards,
> > Alistair+
> >
> >
> >
>
>
>
>
Other related posts:
