On 23/03/2011 1:16 AM, Karsten Bräckelmann wrote:
On Wed, 2011-03-23 at 00:32 -0400, Daryl C. W. O'Shea wrote:
On 23/03/2011 12:27 AM, Karsten Bräckelmann wrote:
Also, while this Perl bug is out there (which undoubtful will exists for
a long time on production machines), should we refrain from tflags
multiple body rules?
Guess we cannot push them, unless guarded by a version 3.3.2 if-block.
If that turns out to be what should be done, lint should be updated to
catch when its not done so that it breaks the build and prevents the
rules from ending up in an update to old versions.
Clearly, you are speaking about lint checking on our British Butler,
aren't you?
Um, sure. If trunk lint checks for the requirement we'll be sure that
such ifversion lines are included for new rules (and old rules,
actually) in sandboxes. If the ifversion lines are forgotten lint will
fail, mass-checks will break, and thus no scores will be generated and
no updates will be released. In this way, making trunk lint for stuff
really does give you protection against old versions.
There's no need to have the old versions lint for things if trunk
prevents us from releasing rules with the stuff in them.
Of course, people are on their own for their own locally crafted rules
(running already released code).
Daryl
I mean, we cannot have old SA versions lint check for this. If we could,
we just could have fixed it instead. Which we cannot.
Though yeah, agreed, a lint check at least for Jenkins to catch
inappropriate tflags multiple for body rules and version<3.3.2 sounds
like a good precaution. Might back-fire, though, due to generated
scores. In the worst case, this is scorched earth until 3.4. :(
