On Sat, Mar 06, 1999 at 09:41:35AM +0000, Jules Bean wrote: > > > > I usually extract the new one (or apply a patch if that is the method), > > > > gunzip -dc old_version.diff.gz > somefile, then edit that 'somefile' and > > > > replace all instances of progdirectory-1.2.0 with progdirectory-1.2.1. > > > > Then apply that new diff with patch -p0 < somefile. And then clean up > > > > the > > > > .rej and .orig files and apply the missing stuff manually. > > > > > > The editing isn't necessary. > > > > > > Simply patch -p1 < somefile and it will ignore the first directory > > > component (the versioned directory). > > > > > I never had the guts to do that. What happens if the upstream author > > heavily modifed > > a file that is modified in your diff? I guess patch is supposed > > to be smart, but is this failsafe? > > If patch gets confused, it won't apply that patch. > > Sometime's you'll have to work out what's going on by hand. It's fairly > subjective whether it's more hassle to evaluate the patches one-by-one, > and apply those still relevant, or to apply them all and then evaluate > them.
To be more precise, patch will apply hunks that he can - and those that it doesn't manage, will save in <patchedfilename>.rej file. Also, if it has any problems with patching (that includes the 'fuzz', offset of lines), it'll save the original file in <patchedfilename>.orig. I see no way you can lose data. Except if the patch is not malformed, but simply wrong :) BTW Jules, I've tried the -p1 method, it works very well. Thanks for the tip. -- enJoy -*/\*- http://jagor.srce.hr/~jrodin/
