Hi,
(doh… I thought I had sent this mail 7 days ago…)
On Sat, Oct 29, 2016 at 03:48:11PM +0100, James Clarke wrote:
> unless priorities are manually configured locally. The attached patch
> should fix this (compiled but not tested).
Thanks for the patch, but… (continued inline).
> Commit 8279242 also broke the documentation build, so I've attached a
> fix for that as well.
Thanks! Applied and pushed [on 2. Nov] :)
> diff --git a/ftparchive/writer.cc b/ftparchive/writer.cc index
> index 018cf00..0bfe6ba 100644
> --- a/ftparchive/writer.cc
> +++ b/ftparchive/writer.cc
> @@ -1019,7 +1019,9 @@ ReleaseWriter::ReleaseWriter(FileFd * const GivenOutput, string const &/*DB*/) :
> Fields["Signed-By"] = "";
> if (_config->FindB("APT::FTPArchive::DoByHash", false) == true)
> Fields["Acquire-By-Hash"] = "true";
> -
> + Fields["NotAutomatic"] = "";
> + Fields["ButAutomaticUpgrades"] = "";
> +
> for(map<string,string>::const_iterator I = Fields.begin();
> I != Fields.end();
> ++I)
… that works I am sure, but it feels like the wrong interface – after
all: The only reasonable setting is "yes", anything else is bogus. In
sofar I guess I would prefer a boolean config option defaulting to false
and emitting the field with the right value on true instead of being
string based (with the bonus that apt accepts yes/no, 1/0 and a few more
as boolean values, too. People might get "confused" by this and set the
string-based value to true with unexpected results…).
A test would be nifty, too. As noted in an earlier mail our testing
framework adds the field via sed at the moment. It shouldn't be hard to
change it to use the new setting instead. No need to come up with a new
testcase here to be able to confidently say "well tested". :)
Thanks again for working on that issue!
Best regards
David Kalnischkies
Attachment:
signature.asc
Description: PGP signature