[cvsnt] Re: [cvsgui] Re: Unable to correctly import *.tar.gz archives

[Mike Ayers]

Arvind Raman wrote:
> Hi there,
> This might sound interesting- probably might throw out
> a bug
> 1. I tried adding the file (*.tar.gz) to the database
> - checked out the file and compared it with the one
> that I has checked in. It was same.
> 2. But when I tried importing the same file - CVS
> recognized it as a Unicode file - I asked it to force
> it as a binary file (used the -kb option). After the
> import happened I looked at the RCS file and it was
> just a 1KB big. Obviously the file on the database was
> not similar to the one I had asked it to import.
> 
> So there exists some difference between import and add
> even when both of them treat the file in the same way.
> Is that normal ?

	This is the import problem of CVS.  The import command treats all 
files as text files, and there is no way to change this (it ignores 
"-kb").  It seems that at a certain point in your .tar.gz file there 
is a ^Z character, which is the old DOS convention to end a text file, 
and this will cause the read of the file to stop, which is why the 
imported file is so short.

	Now is the part where I tell you "don't do that".  The import command 
is meant to bring in large trees of source code, which would be all 
text files.  If your TGZ file contains all text files, you can expand 
it and import those, otherwise just use `cvs add` to add a single file.


/|/|ike