Re: normalizing files and old revisions

[Andreas Klauer]

On Monday 01 September 2003 19:30, Mike Castle wrote:
> Andreas Klauer  <address@hidden> wrote:

>   -w  --ignore-all-space  Ignore all white space.
>   -b  --ignore-space-change  Ignore changes in the amount of white space.
>   -B  --ignore-blank-lines  Ignore changes whose lines are all blank.
>   -I RE  --ignore-matching-lines=RE  Ignore changes whose lines all match
> RE.

I know these options, but they don't quite do what I want. I do not want to 
completely ignore whitespace changes - I want to ignore spaces <-> tabs and 
LF <-> CRLF conversions, trailing spaces, etc. And my filter may do even more 
than that. These options just don't do what I want.

> One thing I strongly feel about history is, it should never lie.  That's
> how they were checked in.  I may not like them, but I'll leave them be.

Alright, alright, I see your point. But how about I just copy the old 
revisions, filter them, and then put them in a side branch starting at the 
very first (ancestor) revision? This way, all revisions (filtered or not) 
would still be accessible via CVS.

The new revision structure would then look like this:
#A#--->#R#--->...--->#O#
 |   ^ old, unfiltered branch ^
 |
 v   v new, filtered branch v
[A]--->[R]--->...--->[O]--->N-->...--->H

A: Ancestor | R: Revision | O: Old Head | N: New Revisions | H: Current Head
Letters marked with ## means unfiltered, [] is their filtered counterpart.
New revisions and current head are always filtered.

This way, the 'lie' would be reduced to keeping the original dates, logs, 
authors for the new, filtered revisions, because I don't want that data lost.
I'd like to have a connection from #O# to [O], too, (or connections between 
filtered and unfiltered counterparts in general, just to show they are 
related) but I'm not sure that's allowed. The tree example in 'man rcsfile' 
does not contain such connections.

What do you think about it?

Regards,
Andreas