Yeah, subclassing sounds the way forward. I was thinking something like: SignedIntegerField = IntegerField(min_value=–2147483648, max_value= 2147483647) UnsignedIntegerField = IntegerField(min_value=0, max_value=4294967295)
Do you think there would be much chance in having these two field types included in the core? If so, should I create a patch for fields.py and submit?? As for the MySQL strict mode, I'll submit a documentation patch, and see if it gets accepted :) On Sun, May 1, 2011 at 11:58 AM, Mathieu AGOPIAN <mathieu.agop...@gmail.com>wrote: > For reference, this discussion is linked to > > http://groups.google.com/group/django-developers/browse_thread/thread/f0b8ddbda03a2d8e > and to the ticket http://code.djangoproject.com/ticket/15923 > > Cal, > > The only thing that i could imagine regarding "fixing" this issue > would indeed be a documentation addition in the MySQL (or MySQLdb) > part of the http://docs.djangoproject.com/en/dev/ref/databases/ page, > explaining the possible issue with integers that are too large to fit > in the appropriate mysql column (though i'm not sure how all this > works, as i'm far from a MySQL or ORM expert). > Specifically, explaining how to set the sql_mode to "traditional" as > explained by kmtracey (http://code.djangoproject.com/ticket/ > 15923#comment:10) I believe. > > Let's see if there's anyone reading this mailing list showing interest > about this, and/or feel free to create a feature request on trac, and > see if it's accepted. > > On a side note, as python doesn't seem to have any issue with large > integers, i guess you could subclass the IntegerField, and add to it's > validation a check to see if the resulting integer can be stored in a > 32 bits using either a simple comparison, like the following: > -int('1' * 31, 2) < int(field_value) < int('1' * 31, 2) > > Mathieu > > On Apr 30, 8:31 pm, "Cal Leeming [Simplicity Media Ltd]" > <cal.leem...@simplicitymedialtd.co.uk> wrote: > > Hey Mathieu, > > > > Thanks for taking the time to reply. I'm starting to see now why the core > > devs are reluctant to modify IntegerField. > > > > I'm wondering if maybe Django should have a SignedIntegerField and > > UnsignedIntegerField as part of the core (for those that wish to have > > enforced 32-bit integers), with the same INT_MIN and INT_MAX from > limits.h (http://en.wikipedia.org/wiki/Limits.h). But there again, would > this be > > considered un-pythonic or against the ethics of Django? > > > > I guess really it should be up to MySQL to have strict mode by default. > But, > > as this is unlikely to happen, could we perhaps consider having a > commented > > out entry in the settings.py file that allows you to set strict mode for > all > > SQL connections? Or, perhaps a documentation change, which explains > easily > > to the user how to do it (Kinda like the storage_engine thing onhttp:// > docs.djangoproject.com/en/dev/ref/databases/#creating-your-tables) . > > > > Let me know your thoughts :) > > > > Cal > > > > On Sat, Apr 30, 2011 at 6:32 PM, Mathieu AGOPIAN > > <mathieu.agop...@gmail.com>wrote: > > > > > > > > > > > > > > > > > Hello, > > > > > I'm afraid there isn't such a thing as "a valid signed value", if we're > > > still talking about "size wise". > > > > > For django (python), the integer you gave in the ticket is perfectly > valid. > > > Here's a way for you to check that : > > > >>> s = '351760125423456632454565345363453423453465345453' > > > >>> int(s) > > > 351760125423456632454565345363453423453465345453L > > > > > And indeed, an IntegerField validates that the content of the field can > be > > > converted to an int this way (check django/forms/fields.py line 230). > > > > > So definitely, as Alex pointed, this is an issue on MySQL's side, not > > > Django's. > > > I believe this can't (shan't?) be fixed at Django's level, as there's > no > > > "size" limitation for the IntegerField, as you would have on a > CharField > > > with the *max_length* attribute. > > > > > And no, limiting the length of the string won't work, as "2147483647" > isn't > > > the same length as "-2147483647", but is the same length as > "9999999999" (if > > > we're taking the example of 2^32-1 as the max SIGNED INT value). > > > > > my two cents ;) > > > > > -- > > > You received this message because you are subscribed to the Google > Groups > > > "Django developers" group. > > > To post to this group, send email to > django-developers@googlegroups.com. > > > To unsubscribe from this group, send email to > > > django-developers+unsubscr...@googlegroups.com. > > > For more options, visit this group at > > >http://groups.google.com/group/django-developers?hl=en. > > -- > You received this message because you are subscribed to the Google Groups > "Django developers" group. > To post to this group, send email to django-developers@googlegroups.com. > To unsubscribe from this group, send email to > django-developers+unsubscr...@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/django-developers?hl=en. > > -- You received this message because you are subscribed to the Google Groups "Django developers" group. To post to this group, send email to django-developers@googlegroups.com. To unsubscribe from this group, send email to django-developers+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/django-developers?hl=en.
