On 06/07/11 07:20, Wieland Hoffmann wrote:
Hallo, Allan McRae:
On 04/07/11 22:13, Wieland Hoffmann wrote:
Looking good. Some general comments:
I saw that --skipinteg implies --skippgpcheck. I noticed this when
I copied a "bad" signature into my source directory and I did not
update the md5sums so used --skipinteg. I was quite surprised when
the signatures did not get checked. Should these be separated more?
I chose to implement it this way because checking the signature means
verifying that the data I downloaded is the data uploaded by the
project which is what data integrity is about. Personally, I would be
surprised if --skipinteg didn't imply --skippgpcheck, although it's kind
of doing the same thing twice. Maybe a switch like --skipchecksums would
be a good idea that doesn't imply skipping ALL integrity checks.
That sounds a good idea
--skipinteg - skips all
--skipchecksums - skips checksums only
--skippgpcheck - skips pgp sig check only
Leave the current patch as is with respect to how this is handled and
add that in a separate patch (if you want to do that work...)
+ local file
+ local errors=0
We should keep track of the number of non-error warnings too so a
"==> WARNING:" message could be outputed.
The exact number/reason or just a simple "hey, there were some warnings"
so people scroll up to the actual warning(s)?
Just a "hey there were some warnings" is enough.
Allan
