Re: Possible Django bug I am considering raising

['OwlHoot' via Django users]

On Thursday, 4 June 2020 16:25:09 UTC+1, OwlHoot wrote:
>
>
> I believe the issue I have enquired about on the following Stack Exchange 
> page :
>
>
> https://stackoverflow.com/questions/62162288/baffling-error-filtering-django-objects-of-derived-class-by-values-in-foreign-ke
>

By adding some debug code in 
/usr/lib/python3.6/site-packages/django/db/backends/base/creation.py I have 
made some progress with this "bug" (which I am no longer convinced it is).

The issue relates to a load of derived classes, as follows :

class DataSet(models.Model)

class PSessionDataSet(DataSet)
   :::
class FilteredDataSet(DataSet)
class AggregateDataSet(DataSet)

   class SurveyRunDataSet(DataSet)

   class RestrictedDataSet(SurveyRunDataSet)

These all have the same app_label value ("dataset") in their Meta class, 
and are thus in the same "group" of models.

However, the debug logging line reveals that Django thinks all these 
classes share the same default manager (_default_manager), namely 
dataset.AggregateDataSet.objects:

Model   
 Default Manager
 :::
app.survey.models.grid.GridRow   
survey.GridRow.objects
app.analysis.models.Analysis 
analysis.Analysis.objects
app.analysis.report.models.AnalysisReport   
 report.AnalysisReport.objects
app.analysis.dataset.models.dataset.DataSet 
 dataset.AggregateDataSet.objects
app.analysis.dataset.models.survey_run.SurveyRunDataSet 
 dataset.AggregateDataSet.objects
app.analysis.dataset.models.filtered.FilteredDataSet 
dataset.AggregateDataSet.objects
app.analysis.dataset.models.restricted.RestrictedDataSet 
dataset.AggregateDataSet.objects

Now I know the documentation says the default manager is chosen as the 
first it happens to come across in the model group. But that sounds 
absolutely barking mad. Shouldn't each of these models have their own 
model-specific default manager, as all the models preceding the "dataset" 
models do in the above table?!

Otherwise surely the manager will have trouble finding various 
columns/values in a model table other than the one on which it is based, 
which is exactly the error I am seeing!

So now my problem becomes how to tweak each of these "dataset" models so 
their default managers will become as follows :

Model   
 Default Manager
 :::
app.survey.models.grid.GridRow   
survey.GridRow.objects
app.analysis.models.Analysis 
analysis.Analysis.objects
app.analysis.report.models.AnalysisReport   
 report.AnalysisReport.objects
app.analysis.dataset.models.dataset.DataSet 
 dataset.DataSet.objects
app.analysis.dataset.models.survey_run.SurveyRunDataSet 
 dataset.SurveyRunDataSet.objects
app.analysis.dataset.models.filtered.FilteredDataSet 
dataset.FilteredDataSet.objects
app.analysis.dataset.models.restricted.RestrictedDataSet 
dataset.RestrictedDataSet.objects

Would that just be a matter of adding lines as follows (typically in the 
FilteredDataSet class) :

  objects = FilteredDataSet.Manager()

or does one need a more elaborate overriding of get_queryset() ?

Any ideas gratefully received.

?

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/961761ab-10d8-4f7d-bc24-961200f54e5co%40googlegroups.com.

Re: Possible Django bug I am considering raising

[Ramiro Morales]

Remember you can not juat open/report an issue and bisect the git commit
which could have introduced the change in behavior, but also, if it gets
confirmed as a real bug, you can fix it yourself so to not delay yourself
in your current project. With the nice side effect that the fix will
benefit many others.

Another option would be: If you can't or don't want to do what I suggest
above then Are you willing to commit to funding another Django contributor
to get this immediate attention & personalised service so to be able to
solve the issue affecting your current project/client?

On Fri, Jun 5, 2020, 7:50 AM 'OwlHoot' via Django users <
django-users@googlegroups.com> wrote:

>
>
> On Thursday, 4 June 2020 23:45:53 UTC+1, Tim Graham wrote:
>
> You can try to find the commit that introduced the issue:
>> https://docs.djangoproject.com/en/dev/internals/contributing/triaging-tickets/#bisecting-a-regression
>>
>
> I'm happy to try and do this, i.e. produce a simple test case that fails,
> and then find the commit that first caused this, provided I am assured that
> if I can then a Django maintainer/developer will jump on this and fix it
> pronto!
>
> If the result will be added to some list where someone will think about
> maybe fixing it in six months, and then only if anyone else complains, I'll
> have long since moved on to another project and/or another client and the
> issue will be of no interest to me.
>
>
>
>
>
>
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Django users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to django-users+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/django-users/016c2855-ec5e-4880-ac18-c4f5660732dbo%40googlegroups.com
> 
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/CAO7PdF8%3DDfYBbd4UxwvRfdxvgRx9fRp_fm9jLMN1AGVz9YeZjA%40mail.gmail.com.

Re: Possible Django bug I am considering raising

[René Fleschenberg]

Hi,

the code you posted on stackoverflow is not sufficient to reproduce the
problem. It would be good to provide a minimal reproducible example (see
also https://stackoverflow.com/help/minimal-reproducible-example).

Regards,
René

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/85bf94fa-d3ae-6b10-dae4-6c52aa58b3c3%40fleschenberg.net.

Re: Possible Django bug I am considering raising

['OwlHoot' via Django users]

On Thursday, 4 June 2020 23:45:53 UTC+1, Tim Graham wrote:

You can try to find the commit that introduced the issue: 
> https://docs.djangoproject.com/en/dev/internals/contributing/triaging-tickets/#bisecting-a-regression
>

I'm happy to try and do this, i.e. produce a simple test case that fails, 
and then find the commit that first caused this, provided I am assured that 
if I can then a Django maintainer/developer will jump on this and fix it 
pronto!

If the result will be added to some list where someone will think about 
maybe fixing it in six months, and then only if anyone else complains, I'll 
have long since moved on to another project and/or another client and the 
issue will be of no interest to me.






 

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/016c2855-ec5e-4880-ac18-c4f5660732dbo%40googlegroups.com.

Re: Possible Django bug I am considering raising

[Tim Graham]

You can try to find the commit that introduced the issue: 
https://docs.djangoproject.com/en/dev/internals/contributing/triaging-tickets/#bisecting-a-regression

On Thursday, June 4, 2020 at 11:25:09 AM UTC-4, OwlHoot wrote:
>
> Hi all
>
> I believe the issue I have enquired about on the following Stack Exchange 
> page :
>
>
> https://stackoverflow.com/questions/62162288/baffling-error-filtering-django-objects-of-derived-class-by-values-in-foreign-ke
>
> may well be a Django bug, or at the least no longer works with Django as 
> this has evolved since v 1.10 (when it worked fine).
>
> In summary, the issue is that if a model class B inherits from A, which 
> inherits from models.model, then a call to B.objects.filter() cannot locate 
> a field actually defined in B (because, it seems, Django starts its walk 
> through the relevant/related models from A rather than B and for some 
> unfathomable reason checks everything _except_ B !)
>
> Regards
>
> John R Ramsden
>

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/ce8987ee-a001-47bd-8d74-2d1b8469da25o%40googlegroups.com.

Possible Django bug I am considering raising

['OwlHoot' via Django users]

Hi all

I believe the issue I have enquired about on the following Stack Exchange 
page :

https://stackoverflow.com/questions/62162288/baffling-error-filtering-django-objects-of-derived-class-by-values-in-foreign-ke

may well be a Django bug, or at the least no longer works with Django as 
this has evolved since v 1.10 (when it worked fine).

In summary, the issue is that if a model class B inherits from A, which 
inherits from models.model, then a call to B.objects.filter() cannot locate 
a field actually defined in B (because, it seems, Django starts its walk 
through the relevant/related models from A rather than B and for some 
unfathomable reason checks everything _except_ B !)

Regards

John R Ramsden

-- 
You received this message because you are subscribed to the Google Groups 
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to django-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/django-users/f694d71d-bb1a-4e60-80ff-c8589f797f51%40googlegroups.com.