On Tue, 18 Jun 2002, Joerg Schilling wrote:
>
> >From [EMAIL PROTECTED] Tue Jun 18 19:54:13 2002
>
> >> People do not have interest because they don't know and this is because VENDORS
> >> do not put star on the distributions.
>
> >It's not just Linux and BSD (and HURD?), I don't know of anyone who even
> >provides it, much less makes it default.
>
> This is _not_ the problem. The real problem is that there _are_ vendors who
> make GNU tar the default and lack hints for alternatives.
If every vendor on the planet chooses not to include your software, I
would suggest you at least consider that maybe the problem is that your
software doesn't suit their needs. As in they need to make the customers
happy by conforming to the de-facto standard instead of the ANSI standard.
> >That is exactly the point I was making. Star is better at doing things
> >most people don't do. Therefore "GNU tar is sufficient," at least for most
> >people.
>
> Star is doing a lot of things in a nice and easy way that could not or only be
> done with a lot of problems whan not using star. This is not only archiving...
> Star is e.g. the only tool on UNIX that is able to 100% compare two file trees
> in only one run.
Then you have to assume that the vast majority of people don't do that...
> >> Instead for unknown reasons, vendors put the non-standard compliant GNu tar on.
The reasons are known, you just don't accept them.
1. star creates tar files which don't always work right with GNU tar
2. most people use GNU tar or vendor tar and need a format they can read
3. most people have zero need or even desire for the features beyond xct
4. almost everyone has scripts which use tar and star uses different
command line syntax and won't work
> Unfair is that FSF does not like GNU tar to be standard compliant. They are
> using the same methods as M$ does. The best idea is to boycot programs that are
> trying to force other people to use their software by widely spreading
> non-standard programs.
FSF is a "vendor" for the purposes noted above, works for most users if
more important than complies with a standard people aren't currently
using.
> I did send a long list of bugs for GNU tar to the maintainers in 1994!
> I even promised them a way to help them to migrate to be standard comforming
> without problems.
As noted, the de-facto standard is widely used, ANSI isn't. Life isn't
fair. By default GNU should read and understand ANSI (and I have a note
from someone at FSF saying they comply with the standard as of 1998 or
so). It would be nice to write ANSI as an option, but few people care or
they would use cpio, star, whatever.
> If they did listen to me in 1994, GNU tar would now be able to understand that
> there is a difference between real TAR archives and GNU tar archives.
I'm sure you used your well-known tact to say this...
> GNU tar
> would not assume that everything is a GNU tar file even when it clearly looks
> different. GNU tar could be able to write POSIX TAR files by request for 8
> years. If now somebody said: let us switch create POSIX tar files by default.
By option would be good. By default I doubt.
> But this is the big problem I also have to critisize with Linux. There is
> nobody able/willing to understand that there is (or soon will be) a problem
> and that there should be a migration path to a better solution!
Why do you not say this about your favorite Solaris? Or AIX, UniCOS, etc?
They don't use star either?
-- rob bogus
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
