1549: The email-address 'myname@localhost' is not considered valid.

holmstro********@gmai***** (Google Code) (Is this you? Claim this profile.)
June 10, 2010 
What version are you running?
1.0.5.1

What's the URL of the page containing the problem?
/account/preferences/

What steps will reproduce the problem?
1. Login for the first time on a fresh installation
2. enter 'myname@localhost' as email address
3. submit

What is the expected output? What do you see instead?
I expect my account settings to be updated, instead I get the message
'Enter a valid e-mail address.'

What operating system are you using? What browser?
Linux vurt 2.6.31-20-generic #57-Ubuntu SMP Mon Feb 8 09:05:19 UTC 2010
i686 GNU/Linux
Firefox/3.5.8
Python 2.6.4

Please provide any additional information below.
Since I entered this email address when running rb-site it was accepted
then this is not only a validation problem but also inconsistent behaviour.

Django version installed is Django-1.1.1.
chipx86
#1 chipx86 
This is due to Django's validation. I don't think it's unreasonable to mark that as
an invalid e-mail address, but we should certainly be consistent.

Do you have a strong need for a non-FLQN in the e-mail address?
  • +NeedInfo
  • -Priority-Medium
    +Priority-Low
    +Component-Settings
    +Component-RB-Site
    +Milestone-Release1.5
#2 holmstro********@gmai***** (Google Code) (Is this you? Claim this profile.) 
In general I can sort of live with flawed email-address validation since this is so
common and also understand djangos problem with altering their email-address validator.

That said I think that not allowing localhost is kind of a nuisance when evaluating a
product like reviewboard. During evaluation it would be nice to be able to simply use
localhost to see how email is handled without spamming already existing
email-accounts or setting up a fake host. This, of course, was not a big problem and
you are certainly not the only ones that have it.

I did however find that the inconsistency between the setup-tool(nice work by the
way, very easy to use) and the application made me feel less confident about the
overall quality of reviewboard, something which could easily be avoided.
chipx86
#3 chipx86 
Thanks. That's good feedback. We should make a decision as to how much validation we want to do, 
and where.

Are there other things in the installer that you found to be inconsistent?
#4 holmstro********@gmai***** (Google Code) (Is this you? Claim this profile.) 
I didn't have any other huge problems with the installer but there was another thing
you'd might consider.

Since I was evaluating I sandboxed reviewboard and it's dependencies by using
virtualenv, which made it very hard for me to use my systems apache or lighttpd. This
was something I didn't think about until the setup was done but since lighttpd is
easy to run standalone on any port this wasn't much of a problem, I just had to add
server.document-root (I used /tmp/) and server.port to the generated lighttpd.conf to
make it work(don't know if this will cover all cases).

With all the problems concerning distribution packages and python packages I don't
think I'm the only one that does this and it would've been nice to have some pointers
in the documentation, output from rb-site or/and commented options in lighttpd.conf
to make testing even simpler.
david
#5 david 
It sounds like there's not really anything to do from this. We provide a minimal lighttpd configuration but I don't think we want to document every possible config in there, because the lighttpd docs are much better on the topic.
  • -NeedInfo
    +WontFix