[comp.emacs] Compressed and split gcc's are on prep and ingres.berkeley.edu

gnu@INGRES.BERKELEY.EDU.UUCP (03/23/87)

I have compressed /u2/emacs/gcc.tar to create gcc.tar.Z.  I also split
that down into 100K chunks, named gcc.tar.Z.1 through gcc.tar.Z.11.  If
you rcp or ftp it, please get the split files instead of the original,
since (1) they are less than half the data, and (2) if the Arpanet goes
belly-up midway, you won't have to start from scratch.

In general when snarfing some large thing on prep (or elsewhere), look
for a .Z file; and if one doesn't exist and you have a prep account,
create one.  It will be faster than transferring the uncompressed
file.  It'd be nice if the person creating the distribution would also
create the compressed file, so it wouldn't take another message to the
mailing list to announce it...

I am ftp-ing the split gcc files to the ftp pub directory on
ingres.berkeley.edu, for all you west coast GNU fans.  They will also
be available from hoptoad's /usr/spool/uucppublic, if you have a link
to hoptoad.
 
        John

wohler@sri-spam.UUCP (03/24/87)

In article <8703230652.AA03678@ingres.Berkeley.EDU> gnu@INGRES.BERKELEY.EDU (John Gilmore) writes:
>I have compressed /u2/emacs/gcc.tar to create gcc.tar.Z.  I also split
>that down into 100K chunks, named gcc.tar.Z.1 through gcc.tar.Z.11.

john,

  i'm surprised you would commit two errors here.  first, you *should*
  have placed the split files in a directory called "gcc.tar.Z-split"
  to minimize the number of files in the already too fat emacs
  directory, and secondly, you should have used split(1) to create
  names of the form gcc.tar.Z.aa, gcc.tar.Z.ab, ..., gcc.tar.Z.ba and
  so on.  with your numbering scheme, i couldn't just say
  
	cat gcc.tar.Z.* > gcc.tar.Z

  to put the file back together--i had to enumerate all of the
  extensions.

  the split on prep has a -c option so that you can specify the byte
  count of each file. 

  i cc'd comp.emacs on this so that others would not learn from this
  bad example.  sorry john.

		      					--bw

rbj@ICST-CMR.ARPA.UUCP (03/31/87)

? In article <8703230652.AA03678@ingres.Berkeley.EDU> gnu@INGRES.BERKELEY.EDU (John Gilmore) writes:
? >I have compressed /u2/emacs/gcc.tar to create gcc.tar.Z.  I also split
? >that down into 100K chunks, named gcc.tar.Z.1 through gcc.tar.Z.11.
? 
? john,
? 
?   i'm surprised you would commit two errors here.  first, you *should*
?   have placed the split files in a directory called "gcc.tar.Z-split"
?   to minimize the number of files in the already too fat emacs
?   directory, and secondly, you should have used split(1) to create
?   names of the form gcc.tar.Z.aa, gcc.tar.Z.ab, ..., gcc.tar.Z.ba and
?   so on.  with your numbering scheme, i couldn't just say
?   
? 	cat gcc.tar.Z.* > gcc.tar.Z

I would recommend that we seriously rethink our distribution scheme.
I object to this method on several grounds: [1] given that you had to
explain the method, it is not obvious (maybe you should include a shell
script or makefile to automate splitting/reassembling, [2] even tho
each piece is *transmitted* separately, the whole depends on *every* bit
of information, [3] which makes selection of certain pieces impossible,
[2] huge amounts of space are required for the uncompress.

I propose that each directory be first tarred, and then compressed. If
the resulting file is too large, then limit the tar, as in
lisp.a-l.tar.Z and lisp.m-z.Z. Or perhaps lisp.el.tar.Z and
lisp.elc.tar.Z. After all, must we be forced to retrieve the cpp,
shortnames, and manual directorys every time we snarf emacs? Do we
really need the .elc files or could we do byte-recompile-directory
just as well. And now that gcc is a separate product, shouldn't we
separate gdb as well? And what of all the other futilitys that make up
the gnu project? Where are they? C'mon, let's get it together!

?   to put the file back together--i had to enumerate all of the
?   extensions.
? 
?   the split on prep has a -c option so that you can specify the byte
?   count of each file. 

Which brings me to another point. What is the real limit on file size?
32K? 65K? 100K? 200K? Anything that sounds reasonable?
 
?   i cc'd comp.emacs on this so that others would not learn from this
?   bad example.  sorry john.
? 		      					--bw

	(Root Boy) Jim "Just Say Yes" Cottrell	<rbj@icst-cmr.arpa>
Is he the MAGIC INCA carrying a FROG on his shoulders?? Is the FROG
his GUIDELIGHT?? It is curious that a DOG runs already on the ESCALATOR...