#django

/

      • zamro has quit
      • Debnet has quit
      • ams__ has quit
      • fission6 has quit
      • schinckel
        So I'm having some issues with deadlocks when applying migrations to my Postgres database. The migrations in question are all CREATE TABLE (or CREATE UNIQUE, on these newly created tables).
      • They shouldn't be generating deadlocks, from what I can see.
      • It's not the same queries they are deadlocking with.
      • Has anyone else hit this?
      • jas02_ joined the channel
      • b00tstr4p has quit
      • Has hand-tweaking my migrations so they are all a single CreateModel (instead of having seperate AddColumn for the FKs) helped cause this?
      • mmxx_th has quit
      • FunkyBob
        I read that as "hand-twerking" ... and thought it was some new craze
      • XHFHX has quit
      • schinckel
        rofl
      • kyheo joined the channel
      • So my two planned approaches are to (a) just regenerate the migrations using "makemigrations", and (b) split the migration file into one per operation.
      • I think the former is nicer, so might try that first.
      • keimlink joined the channel
      • zamro joined the channel
      • keimlink_ has quit
      • CrowX- has quit
      • Darkhunter has quit
      • jas02_ has quit
      • goldfish joined the channel
      • renlo joined the channel
      • Bryson joined the channel
      • shangxiao joined the channel
      • So, it seems that my combining the AddFields was at least partly complicit.
      • Regenerating the migrations worked first time.
      • Darkhunter joined the channel
      • Lesson learned. Blog post likely.
      • FunkyBob
        mmm?
      • Darkhunter has left the channel
      • schinckel
        So I was making nice "clean" migrations by moving the AddFields that created FKs into the CreateModel operations.
      • Seems that maybe it's a deliberate decision to _not_ do that.
      • FunkyBob
        ah
      • schinckel
        I (mistakenly) thought that doing that would make the migrations apply faster and less chance of deadlock.
      • But it seems otherwise.
      • FunkyBob
        MarkusH might know... or andrew
      • woltman joined the channel
      • schinckel
        Yeah. It was MarkusH's blog posts that emboldened me to twerk (!!!) them. So I'll blame him ;)
      • mattmcc
        Deferring FK constraints is convenient in most cases, but required in others (circular references)
      • schinckel
        I thought it was just a side-effect of the CreateModel migration generation happening in alphabetical order.
      • mattmcc
        I think it's more like they get to create them alphabetically because they know the constraints come later.. :)
      • schinckel
        Perhaps.
      • mattmcc
        I think South might've tried to create tables in the 'right' order and only defer when necessary, but it's been a while.
      • I might also be thinking of pre-South syncdb. :/
      • Electrometro joined the channel
      • schinckel
        Hahaha.
      • fission6 joined the channel
      • winter98 joined the channel
      • mucco joined the channel
      • renlo joined the channel
      • fission6 has quit
      • mucco_ has quit
      • FancyCamel has quit
      • winter98 has quit
      • sol1x has quit
      • narfman0 joined the channel
      • zivester joined the channel
      • renlo joined the channel
      • donnex has quit
      • thinkt4nk joined the channel
      • zivester has quit
      • c17r joined the channel
      • MarkusH
        schinckel: wait, what?
      • schinckel
        I'm teasing, a little.
      • toc has quit
      • MarkusH
        schinckel: no, srsly, details please
      • schinckel
        I ran into an issue where altering migrations (to have fewer operations, by inlining the foreign key fields into the CreateModel statements) resulted in deadlocks.
      • ticketbot has quit
      • MarkusH
        details
      • what models did you have
      • ticketbot joined the channel
      • what were your initial migrations and how were the once you optimized?
      • schinckel
        I might have to PM you the differences in those.
      • Details will come.
      • Just in the middle of an email.
      • MarkusH
        k
      • if you can drop me an example I'll take a look.
      • But will go to sleep now
      • schinckel
        Yeah, will do.
      • Rialynn joined the channel
      • holler joined the channel
      • holler
        hey, is there an easy way to enable plain text password in django admin for user?
      • as opposed to the hash form
      • FunkyBob
        holler: no?
      • there is no way, since plain text passwords are never stored
      • schinckel
        You really don't want that.
      • holler
        FunkyBob: I have an admin user that needs to create some other user accounts but he cant enter a plain password for them
      • what do you suggest
      • FunkyBob
        holler: whoever it is didn't read
      • Rialynn
        Hi - I'm implementing Django's password reset and I want to customize the html for them but for some reason it seems that it always wants to use the default html. I tried to create a directory called admin inside of my templates directory and i have my app above the django.contrib.auth in INSTALLED_APPS in settings.py but it still only uses the default html. Any ideas?
      • FunkyBob
        well, I've frequently created new users, and admin lets you enter plain passwords on creation
      • or also offers a password change form
      • schinckel
        holler: You can _enter_ a password, but it's _never_ stored in plain text.
      • holler
        hm
      • mrkent joined the channel
      • Koterpillar
        Rialynn: so which template did you override?
      • holler
        when I try, I set password "temp1234" and click save, on page refresh it still shows "password: temp1234"
      • dj 1.9
      • er 1.8
      • FunkyBob
        holler: READ THE FINE PRINT... under that field it tells you you can't enter the password there
      • and then gives you a link to the password change form
      • holler
        FunkyBob: there is no fine print!!! maybe its bc we created custom user
      • that's probably why
      • yeah we're missing that
      • Rialynn
        @Koterpillar well ideally all of them but specifically i want to modify the html for password_reset_compete.html
      • amcorreia has quit
      • FunkyBob
        "Raw passwords are not stored, so there is no way to see this user's password, but you can change the password using this form.
      • holler
      • FunkyBob
        "
      • Koterpillar
        Rialynn: that template is registration/password_reset_complete.html, not admin/password_reset_compete.html
      • Rialynn: you need myapp/templates/registration/password_reset_compete.html
      • schinckel
        holler: Is that a custom form?
      • holler
        schinckel: ya I think
      • jas02_ joined the channel
      • schinckel
        You might want to look at how the auth forms/views work, and make sure you do the same thing.
      • You can usually just subclass the relevant auth form.
      • holler
      • I see.. so you mean subclass some existing auth modeladmin for user?
      • Rialynn
        @Koterpillar That was it. Thank you. I think I had just confused myself a bunch.
      • schinckel
        holler: Yes.
      • jas02_ has quit
      • thinkt4n_ joined the channel
      • lacrymology joined the channel
      • zivester joined the channel
      • zoolook has quit
      • hipertracker joined the channel
      • the_rat joined the channel
      • dardevelin joined the channel
      • microdex has quit
      • isthmian joined the channel
      • blablupp has quit
      • renlo joined the channel
      • the_rat has quit
      • hipertracker has quit
      • hipertracker joined the channel
      • waterCreature joined the channel
      • scagnetti joined the channel
      • kyheo has quit
      • iiie has quit
      • lacrymology has quit
      • lacrymology joined the channel
      • iiie joined the channel
      • iiie has quit
      • slav0nic has quit
      • holler has quit
      • Rialynn has quit