On Sun, Jun 27, 2021 at 12:18 PM David Bremner wrote:
>
> Felipe Contreras writes:
>
> > That way we don't need pass them to the query object ourselves.
>
> I have applied this change to master. As we discussed previously, I
> prefer to leave the sort order explicit in the tests.
OK. I still maintain that I see no point in doing this.
Just to find out what would happen if in the future the default
changed, I used NOTMUCH_SORT_UNSORTED instead of
NOTMUCH_SORT_NEWEST_FIRST in just two places and the result is 32
tests failing.
That does *not* include T395-ruby.sh.
In order for the ruby tests to fail the default in notmuch-search.c
has to be different from the default in lib/query.cc--I don't know why
notmuch-search.c doesn't use the true default.
But these are the tests that are *already* in the master branch. My
changes would not make the situation significantly worse.
I think if we really wanted to make the tests (and the binaries) truly
sort-agnostic, a lot more work is needed--and that's mostly orthogonal
to the changes I proposed to the ruby tests.
diff --git a/lib/query.cc b/lib/query.cc
index 792aba21..b618bf42 100644
--- a/lib/query.cc
+++ b/lib/query.cc
@@ -107,7 +107,7 @@ notmuch_query_create (notmuch_database_t *notmuch,
query->query_string = talloc_strdup (query, query_string);
-query->sort = NOTMUCH_SORT_NEWEST_FIRST;
+query->sort = NOTMUCH_SORT_UNSORTED;
query->exclude_terms = _notmuch_string_list_create (query);
diff --git a/notmuch-search.c b/notmuch-search.c
index 244817a9..83329a4d 100644
--- a/notmuch-search.c
+++ b/notmuch-search.c
@@ -770,7 +770,7 @@ _notmuch_search_cleanup (search_context_t *ctx)
static search_context_t search_context = {
.format_sel = NOTMUCH_FORMAT_TEXT,
.exclude = NOTMUCH_EXCLUDE_TRUE,
-.sort = NOTMUCH_SORT_NEWEST_FIRST,
+.sort = NOTMUCH_SORT_UNSORTED,
.output = 0,
.offset = 0,
.limit = -1, /* unlimited */
--
Felipe Contreras
___
notmuch mailing list -- notmuch@notmuchmail.org
To unsubscribe send an email to notmuch-le...@notmuchmail.org
