Mixing populated query results with model objects

[DaBlick]

I have a Schema sorta like this (abbreviated here):

var Comment = module.exports = new Schema({
  , teacher :   { type: Schema.ObjectId, index: true, ref: 'User'}

var StudentProfile = module.exports = new Schema({
  , comments : [Comment]

Here's my code to add a new comment. You can presume that std contains the result of a query like

       db.model('StudentProfile').findById(id).populate("comment")

And now I execute this code:

        var comment = new Comment;
        comment.teacher    = ~~~ some User object

        std.comments.push(comment);

        std.save(function (err) {...}

        res.render('reporting/_comment', {
            'comments' : std.comments,

Assume there is already one Comment and I'm pushing a second one.

Here's what std.comments look likes at the render call

   [ {teacher : {User object}},
     {teacher : id of User object} }   <--- Newly pushed comment

What you can see here is that the pre-existing Comment from the query result is populated, but the new one is not.

With the advent of the populate feature this kind of scenario is not going to be uncommon and it would be quite nice if the model objects the mongoose magic that turns comment.teacher back into an id was transparent.

[ironstrider]

I've encountered this issue as well, and it is mighty frustrating. Any word on a fix?

[aheckmann]

relates to #601

[pykler]

If there was a manual_populate method, it might solve some of these issues ... another solution would be instead of having:

comment_obj.teacher = teacher_obj

Automagically just keep the teacher_obj._id and leave the whole obj ... it would keep the full teacher_obj but when saving checks the schema and then only saves the teacher_obj._id at that point.

[braunsquared]

+1 Needs to be consistent. Either store the field as an id always or an object. Preference would be an id, with a call to a populate like call that returns the object if present.

[aheckmann]

fixed in 3.6