to file-based namespaces. Confirmed and QAed these are only for
file-based. This moves the majority of graph-parser/parse-file calls
into file-based namespaces. The only ones left are in handler.file which
can be done later
A new existing page was silently being overwritten by a schema.org page
Added detection for future conflicts with existing pages as well
as conflicts of renamed pages. Renamed conflicting properties
in same way as classes. Also added a debug option to create an edn
file which is idempotent and helpful to confirm there are no
regressions
- Reverted config/db-based-graph? check in a db only namespace
- Removed unused ns as it appears we just used another fn instead.
- /cc @RCmerci on failing test
- bb test:load-all-namespaces-with-nbb now passes for all deps
- Updated nbb-logseq to support cljs.core/Atom
- Adjusted datascript.impl.entity with previous datascript.db require
so that correct loading occurs in nbb
- fixed missing deps in logseq.outliner
- Converted a minor cljs.spec to malli since malli is nbb compatible
- Changed exists? to some? since they are functionally equivalent for
the given arg and some? is nbb compatible
- remove sqlite-db/close! as its not used and didn't make sense to be
used in other electron ns
- remove sqlite-db/connections atom as its not needed
- remove read-graph as its not necessary
- rename cli.persist-graph as its primary purpose has changed
- update effected scripts
Part of LOG-2953. We shouldn't persist unusuable property type
configurations as shown by this bug. By enumerating what schema
attributes are allowed for each type, we can prevent future bugs
like this. When changing between property types, this also cleans up
:classes, :position and :values that were accidentally hanging around
for certain types. Also modify test since we don't allow users
to use a :default property with :cardinality
re-index a graph should only show up for file graphs.
Also moved db.migrate to file-based since it contains re-index
and multiple file graph query assumptions
