John Ferlan wrote:
On 01/13/2018 02:00 AM, Roman Bogorodskiy wrote:
> Completion in virsh is enabled when readline is available. However,
> when it's not available, we should:
>
> * avoid defining completers with completion functions;
> * in cmdComplete(), mark unused arguments when there's no readline with
> ATTRIBUTE_UNUSED.
> ---
> tools/virsh-domain-monitor.c | 6 ++++++
> tools/virsh-domain.c | 6 ++++++
> tools/virsh.h | 11 ++++++++++-
> tools/virt-admin.c | 14 ++++++++++++++
> tools/vsh.c | 2 +-
> 5 files changed, 37 insertions(+), 2 deletions(-)
>
Any chance to reduce the number of places to change that we can modify
virsh-completer.c to have the "#if WITH_READLINE" where the #else would
just return NULL and uses virCheckFlags(-1, NULL); to avoid the
syntax-check issue for the two API's defined so far?
Reading vshReadlineParse it seems as if the .completer returns NULL,
then it's no different than if the .completer was NULL.
I'm only one cup of coffee in though - so I could be wrong ;-). It
would have been "better" if the two new API's came with some
documentation over usage expectations (grumble, grumble).
John
Sent out v2. Brief testings shows virsh working as usual with this
change.
Roman Bogorodskiy