svn-cur

[Julian H. Stacey]

Hi,
Chris Rees wrote:
> On 27 December 2011 12:37, Julian H. Stacey <jhs at berklix.com> wrote:
> > Hi Stephen
> >> On 12/26/2011 07:55 PM, Julian H. Stacey wrote:
> >> >> Suggestion:
> >> >> Maybe you might want a 5 digit numeric, not 4 digit, like cvs-cur ?
> >> >> (as eg there's often more cvs deltas than other deltas)
> >> >
> >> > Suggestion:
> >> > Also use same convention as before in naming with xEmpty to indicate a base
> >> > ctm (delta) rather than just a differential ctm (delta).
> >> >     mv svn-cur.0001.xz svn-cur.00001xEmpty.xz
> >> >     mv svn-cur.0002.xz svn-cur.00002.xz
> >>
> >> The convention that svn-cur.0001 does not have xEmpty in it is a
> >> convention that has been followed before.  Look at src-9.  It is how the
> >> mkCTM software works.
> >
> > OK, thanks, I had wondered if it was a mistake.
> > I hadn't noticed the example of src-9 till now.
> >
> > The name 'Empty' was a useful marker to distinguish & seperate a
> > full from a differential file....  We're now up to src-9.0077.gz
> >
> > Question:
> > Might there may be trouble when we get to src-9.0100.gz ?
> > A double bid for that file name:
> > - Recipient hosts will want a normal small incremental update.
> > - Sender & ftp mirror servers, that same name needs to hold
> >  2 different sets of content,
> >  - for a differential.
> >  - for a complete archive
> > How will ctm recipients later distinguish which CTM delta to pull via ftp
> > if they both have the same name ? eg
> >        src-9.0300.gz   a complete base CTM archive for later starters
> >        src-9.0300.gz   a differential delta for those ftp'ing after
> >                        maybe a mail outage on recipient loses an odd delta.
> > Using same name in different new directories might also be problematic, as
> > if we had eg
> >        /pub/FreeBSD/development/CTM/src-cur/src-9.0101.gz
> >        /pub/FreeBSD/development/CTM/src-cur/empty-new-dir/src-9.0101.gz
> > all mirror site would need to create directories called /empty-new-dir/
> > which might (I don't know) require permission/manual intervention on mirrors).
> >
> > The scenario I guess hasn't happened before, & doesn't matter when a new
> > archive first starts, as per src-9.0001.gz,
> > but I guess this potential problem approaches as src-9.0077.gz
> > approaches src-9.0100.gz or src-9.0101.gz ?
> >
> 
> Stephen is pointing out that src-9.0001 is not an incremental delta;
> there's no src-9.0000.
> 
> Yes, when it gets to src-9.0100 there'll be an Empty file here,

Ah good, Thanks.

> but
> since there's no difference between 0001 and 0001Empty, there's no
> point in having both files there.

OK.
I'd suggest it's better to rename them with string Empty for consistency ?

PS
  ls -l
  -rw-r--r--  1 mailnull  mailnull  143591995 Oct  3 18:48 src-9.0001.gz
  -rw-r--r--  1 mailnull  mailnull       2516 Oct  3 18:48 src-9.0002.gz
  
  file src-9.000[12].gz
  src-9.0001.gz: gzip compressed data, from Unix, last modified: Sun Sep 25 03:36:30 2011, max compression
  src-9.0002.gz: gzip compressed data, from Unix, last modified: Tue Sep 27 23:15:29 2011, max compression

  file -z src-9.000[12].gz
  src-9.0001.gz: ASCII English text (gzip compressed data, from Unix, last modified: Sun Sep 25 03:36:30 2011, max compression)
  src-9.0002.gz: troff or preprocessor input text (gzip compressed data, from Unix, last modified: Tue Sep 27 23:15:29 2011, max compression)

Cheers,
Julian
-- 
Julian Stacey, BSD Unix Linux C Sys Eng Consultants Munich http://berklix.com
 Reply below not above, cumulative like a play script, & indent with "> ".
 Format: Plain text. Not HTML, multipart/alternative, base64, quoted-printable.
 EU tax to kill London Vetoed   http://berklix.com/~jhs/blog/2011_12_11