[FIX] web_kanban: fix race condition in do_search

The do_search method still had a race condition because the
read_slice method of datasets alter ids.  So, what could happen is
this: a long search is started, then a shorter search is started and
ended before the first search is completed.  The ui is rendered
properly, but then the long search ends and alter the dataset.  As
a result, the Kanban view is broken (cannot open a record anymore
because the dataset has the wrong ids).

This commit fixes the issue by doing basically what the read_slice
method did, except the ids altering part.  However, it introduced
a new problem for x2many kanban views: these kanban view don't have
the same dataset, they have a datasetstatic. The naive substitution
does not work, so I had to reintroduce the old method in X2ManyKanbanView.

0 files changed, 0 insertions, 0 deletions