On Thu, 2023-11-23 at 11:48 +0100, Pablo Neira Ayuso wrote:
> On Wed, Nov 22, 2023 at 07:22:25PM +0100, Thomas Haller wrote:
> > diff --git a/tests/shell/run-tests.sh b/tests/shell/run-tests.sh
> > index 3cde97b7ea17..c26142b7ff17 100755
> > --- a/tests/shell/run-tests.sh
> > +++ b/tests/shell/run-tests.sh
> > @@ -431,6 +431,19 @@ for t in "${TESTSOLD[@]}" ; do
> > elif [ -d "$t" ] ; then
> > TESTS+=( $(find_tests "$t") )
> > else
> > + if [ -f "$t" ] ; then
> > + # If the test name looks like a dumps
> > file, autodetect
> > + # the correct test name. It's not useful
> > to bother the
> > + # user with a failure in this case.
> > + rx="^(.*/)?dumps/([^/]+)\\.(nodump|nft|jso
> > n-nft)$"
> > + if [[ "$t" =~ $rx ]] ; then
> > + t2="${BASH_REMATCH[1]}${BASH_REMAT
> > CH[2]}"
> > + if [ -f "$t2" -a -x "$t2" ] ; then
> > + TESTS+=( "$t2" )
> > + continue
> > + fi
> > + fi
> > + fi
>
> I think it is not worth, users of this infrastructure is very small.
>
> So let's keep back this usability feature for tests/shell.
OK. Fine.
But while closing and dropping it, let me still make the use case
clear.
When I look at a patch in `git-log` and see for example the diff-stat:
src/json.c | 25 +++++++++++++------------
src/parser_json.c | 91 +++++++++++++++++++++++++++++++++++++++++++++----------------------------------------------
tests/shell/testcases/chains/dumps/0021prio_0.json-nft | Bin 61509 -> 61639 bytes
tests/shell/testcases/chains/dumps/0042chain_variable_0.json-nft | Bin 952 -> 1046 bytes
tests/shell/testcases/chains/dumps/0043chain_ingress_0.json-nft | Bin 619 -> 632 bytes
I'd like to re-run those tests. I'd like to do that by copy-pasting the
file names at hand into the terminal.
Thomas
