jnpkrn / clufter / 703 / 6
60%
master: 60%

LAST BUILD BRANCH: next
DEFAULT BRANCH: master
Ran
Files 214
Run time 9s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 59.933% (+0.3%) from 59.61%
MAINT_EXTRA=0

push

travis-ci

jnpkrn 
facts: infer_comp+_cmp_ver: handle tail ver. component(s) missing right

Previously, it could happen that components could get some "extras",
accumulated as traversed through the evolutionary line, that in fact
would not reflect the raw data.  Triggering condition in such case
is version comparison with one the operands consisting of less
version components -- a case in point (not present though, yet!):
corosync 2.4 vs 2.4.3, which would incorrectly make the former
awarded "extras" from the latter.

The apparent solution for that to rectify _cmp_ver function so that
it imagines virtual zeroes at the positions missing for version
components parity between the two operands.  That alone, however,
makes the "facts.TestClusterSystem.test_cluster_pcs_flatiron" unit
test fail starting with "6.5" case.  Now, what becomes a trouble
is that doing so breaks the logic a la "requiring version 1.x,
which can clearly be fulfilled with 1.4" as the comparison runs
as "(1, 0) vs. (1, 4)" rather than "1 vs. (1, 4), i.e., 1 vs. 1"
as before (1.x here implied with how corosync=flatiron is translated
under the hood).  Hence, as the next step, introduce new parameter
at _cmp_ver so as to request just a prefix match comparison, and
apply that where it matters, i.e. directly at the leaf "component"
part of the inference mechanism, infer_comp function.

Signed-off-by: Jan Pokorný <jpokorny@redhat.com>

3584 of 5980 relevant lines covered (59.93%)

0.6 hits per line

Source Files on job 703.6 (MAINT_EXTRA=0)