Hi
On Tue, Mar 7, 2023 at 4:39 PM Markus Armbruster <arm...@redhat.com> wrote:
> marcandre.lur...@redhat.com writes:
>
> > From: Marc-André Lureau <marcandre.lur...@redhat.com>
> >
> > This allows for a more pleasant user experience.
> >
> > Before:
> > $ ./qemu-system-x86_64 -display egl-headless,gl=
> > qemu-system-x86_64: -display egl-headless,gl=: Parameter 'gl' does not
> accept value ''
> >
> > After:
> > $ ./qemu-system-x86_64 -display egl-headless,gl=
> > qemu-system-x86_64: -display egl-headless,gl=help: Parameter 'gl' does
> > not accept value '' (available: 'off', 'on', 'core', 'es')
>
> error.h recommends to user error_append_hint() for this:
>
> * Create an error and add additional explanation:
> * error_setg(errp, "invalid quark");
> * error_append_hint(errp, "Valid quarks are up, down, strange, "
> * "charm, top, bottom.\n");
>
> > Signed-off-by: Marc-André Lureau <marcandre.lur...@redhat.com>
> > ---
> > include/qapi/util.h | 1 +
> > qapi/qapi-util.c | 15 +++++++++++++++
> > qapi/qapi-visit-core.c | 7 +++++--
> > tests/unit/check-qom-proplist.c | 4 ++--
> > 4 files changed, 23 insertions(+), 4 deletions(-)
> >
> > diff --git a/include/qapi/util.h b/include/qapi/util.h
> > index 81a2b13a33..a5363e595d 100644
> > --- a/include/qapi/util.h
> > +++ b/include/qapi/util.h
> > @@ -23,6 +23,7 @@ typedef struct QEnumLookup {
> > } QEnumLookup;
> >
> > const char *qapi_enum_lookup(const QEnumLookup *lookup, int val);
> > +char *qapi_enum_available(const QEnumLookup *lookup);
> > int qapi_enum_parse(const QEnumLookup *lookup, const char *buf,
> > int def, Error **errp);
> > bool qapi_bool_parse(const char *name, const char *value, bool *obj,
> > diff --git a/qapi/qapi-util.c b/qapi/qapi-util.c
> > index 63596e11c5..19ba4b695a 100644
> > --- a/qapi/qapi-util.c
> > +++ b/qapi/qapi-util.c
> > @@ -84,6 +84,21 @@ int qapi_enum_parse(const QEnumLookup *lookup, const
> char *buf,
> > return def;
> > }
> >
> > +char *qapi_enum_available(const QEnumLookup *lookup)
> > +{
> > + int i;
> > + GString *str = g_string_new(NULL);
> > +
> > + for (i = 0; i < lookup->size; i++) {
> > + g_string_append_printf(str, "'%s'", lookup->array[i]);
> > + if (i + 1 < lookup->size) {
> > + g_string_append(str, ", ");
> > + }
> > + }
> > +
> > + return g_string_free(str, FALSE);
> > +}
> > +
>
> If we use error_append_hint(), we can fold it into the reusable helper,
> and name the result error_append_qapi_enum_hint(). Easier to use than
> qapi_enum_available(), because the caller doesn't have to free anything.
>
Why not. That was not my first approach as it makes enumlookup->string
specific to Error, but if they are no other users, that's fair.
>
> > bool qapi_bool_parse(const char *name, const char *value, bool *obj,
> Error **errp)
> > {
> > if (g_str_equal(value, "on") ||
> > diff --git a/qapi/qapi-visit-core.c b/qapi/qapi-visit-core.c
> > index 6c13510a2b..b887cbf3fd 100644
> > --- a/qapi/qapi-visit-core.c
> > +++ b/qapi/qapi-visit-core.c
> > @@ -404,8 +404,11 @@ static bool input_type_enum(Visitor *v, const char
> *name, int *obj,
> >
> > value = qapi_enum_parse(lookup, enum_str, -1, NULL);
> > if (value < 0) {
> > - error_setg(errp, "Parameter '%s' does not accept value '%s'",
> > - name ? name : "null", enum_str);
> > + g_autofree char *avail = NULL;
> > +
> > + avail = qapi_enum_available(lookup);
> > + error_setg(errp, "Parameter '%s' does not accept value '%s'
> (available: %s)",
> > + name ? name : "null", enum_str, avail);
> > return false;
> > }
> >
> > diff --git a/tests/unit/check-qom-proplist.c
> b/tests/unit/check-qom-proplist.c
> > index 79d4a8b89d..eb0b215abb 100644
> > --- a/tests/unit/check-qom-proplist.c
> > +++ b/tests/unit/check-qom-proplist.c
> > @@ -488,8 +488,8 @@ static void test_dummy_badenum(void)
> >
> > g_assert(dobj == NULL);
> > g_assert(err != NULL);
> > - g_assert_cmpstr(error_get_pretty(err), ==,
> > - "Parameter 'av' does not accept value 'yeti'");
> > + g_assert_nonnull(strstr(error_get_pretty(err),
> > + "Parameter 'av' does not accept value
> 'yeti'"));
> >
> > g_assert(object_resolve_path_component(parent, "dummy0")
> > == NULL);
>
> No need to mess with the test then.
>
> What do you think?
>
We hit an interesting limit of the API though, because error_setg() with
&error_fatal will fail immediately before append_hint().
But we can work around it by using a ERRP_GUARD() !
sending v2
