Hello Alex,

Thank you for these updates, the code looks really solid !

Here are a couple of suggestions I have for these change-sets:

For the commit: *bitbake: runqueue, build, dsi: event data change**
***
+            task.save()
+
+        if isinstance(event, bb.build.TaskBase):
+            task.recipe.name = event._package
+            task.recipe.save()

The first save() can go away, as there is another task.save() call a few lines down.

-----

Would it be better to change this:
self._message = "recipe %s: task %s: %s" % (d.getVar("PF", True), t, self.getDisplayName())
into this:
self._message = "recipe %s: task %s: %s" % (self._package, t, self.getDisplayName()) ?

------

identifier = task_information['recipe'].file_path + task_information['task_name']
This may be a very long key. Maybe we can do something like this:
identifier = task_information['recipe'].file_path.split('/')[-1] + task_information['task_name']
This will store only the recipe name+version together with the task_name

If this point is taken, this line:
    identifier = event.taskfile + event.taskname
Should become:
identifier = event.taskfile.split('/')[-1] + event.taskname

! Beware of other lines that may need to change together with this.

=====================================================

For commit:***bitbake: dsi: refactor the BuildInfoHelper code*

-        return machine_info
+        pass

     def create_machine_object(self, machine_information):

Do we need this 'pass' here ?

---

+            task_object.outcome=task_information['outcome']
+
+ task_object.task_executed=task_information['task_executed']

Can we delete the space between these lines ? It makes the first line stand out from the rest and there's no reason for that.

---

+            import traceback

It's best to have all imports at the beginning of the file.

-----

Get task object and get recipe object are not consistent with eachother.
get_recipe should consider the created attribute before trying to store information:

+        try:
+            recipe_object.name=recipe_information['name']
+            recipe_object.version=recipe_information['version']
+            recipe_object.summary=recipe_information['summary']
+ recipe_object.description=recipe_information['description']
+            recipe_object.section=recipe_information['section']
+            recipe_object.license=recipe_information['license']
+ recipe_object.licensing_info=recipe_information['licensing_info']
+            recipe_object.homepage=recipe_information['homepage']
+ recipe_object.bugtracker=recipe_information['bugtracker']
+            recipe_object.author=recipe_information['author']
+        except:
+            pass
+        finally:
+            recipe_object.save()

And also, if we put this in a try except block, maybe it is best to print the traceback then let it go.
We might consider to use a try/except block to the get_task method.

---------

+ for bl in sorted(self.internal_state['build_layers'], reverse=True, key=_slkey):
+            if (path.startswith(bl.layer.local_path)):
+                return bl

Should we add a comment here with what's the motivation behind this ?

--------------

+        identifier = event.taskfile + event.taskname

This also needs to change if we consider my advice on using smaller identifier values, from above.

--------------

+        if isinstance(event, bb.runqueue.runQueueTaskCompleted):
+ task_information['outcome'] = 3 # TODO: needs to use constants

The Django model updates from this patch states the default for this is value 4.
Either this or that need to change for consistency reasons.

--------------

+            e = event
+            e.taskname = pn

Do we need to do this? I think event instead of e makes the code more clear.

------------

+            task_obj = self.orm_wrapper.update_task_object(task_info)

Caution: This method doesn't return anything, so it may be better just to call it, instead of assigning it to an attribute:
        self.orm_wrapper.update_task_object(task_info)

-------------


That's about it :)
I hope this helps!

Thank you,
Calin


On 19.07.2013 01:35, Damian, Alexandru wrote:
Hi guys,

I've performed a deep refactoring of the database interface code.
This was triggered by the opportunity to use dependency information which now we dump from Bitbake to pre-load recipe and task information prior to actual running.

This is fairly deep, as I moved a lot of code around. Because of this, I would like a round of review before merging in the webhob-master. As such, the code has been pushed to:

http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=webhob-poky/master-next
contrib/webhob-poky/master-next

Can you please check this out and let me know if you find something spotty ?

Cheers,
Alex


--
Alex Damian
Yocto Project
SSG / OTC

_______________________________________________
webhob mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/webhob

Reply via email to