yaws

Mikael Karlsson <mikael@...> writes:

> If you get {error,einval} errors in the log it is probably due to the same
> problem as described in another posting:
>
> http://sourceforge.net/mailarchive/forum.php?thread_id=31173398&forum_id=12456
>
> You will have the same problem if you change the yaws.conf file and try
> to hup Yaws, then it will go into spin. (WARNING R11B users).
>
> This was recognised by the OTP/Erlang folks as bug in R11B in the
> gen_tcp module (or the driver it uses), and a fix was expected for
> R11B-3. I haven't been able to verify this yet though.
> Boring bug as it actually makes yapp unusable on R11B, and I need R11B
> for ErlyWeb. Next phase was to try and add stuff so that one could make
> yapps out of ErlyWeb applications.


Hi Mikael, thanks for the response. I do in fact have 11B installed,
as that's what comes with Debian unstable. I've been bitten by the
Yaws hup glitch as well, so tended to just not do that.

What part(s) of ErlyWeb needed R11B? I was thinking I would use the
ErlTL templates, but not necessarily the rest. Since I think yapp
would be quite helpful, maybe I'll downgrade for the time being.

  Cheers,
  Tim

On 2/5/07, Claes Wikstrom <klacke@...> wrote:
>
> Folks, new bugfix release at http://yaws.hyber.org
>
> Lot's of small fixes by tobbe and also some patches collected
> by contributors.

Excellent!

The urlc_total counter was also causing me problems, my application
generates lots of URLs. Good to see this "feature" go away :)

There was a slight compatibility issue... in Yaws 1.65 appmoddata
started with "/" (e.g. "/foo/bar") but in 1.67 they don't (e.g.
"foo/bar").

-bob

Bob Ippolito wrote:
> On 2/5/07, Claes Wikstrom <klacke@...> wrote:
>> Folks, new bugfix release at http://yaws.hyber.org
>>
>> Lot's of small fixes by tobbe and also some patches collected
>> by contributors.
> 
> Excellent!
> 
> The urlc_total counter was also causing me problems, my application
> generates lots of URLs. Good to see this "feature" go away :)
> 
> There was a slight compatibility issue... in Yaws 1.65 appmoddata
> started with "/" (e.g. "/foo/bar") but in 1.67 they don't (e.g.
> "foo/bar").

Ouch - not ok. Will fix immediately. I'm actually getting
pretty annoyed with the way the appmod path has changed over
releases - I'm not blaming - it's probably my fault .... maybe.

Anyway - this is precisely the sort of thing a decent tests suite
would cover. Any takers !!!!!


/klacke

Claes Wikstrom wrote:

> Ouch - not ok. Will fix immediately. I'm actually getting
> pretty annoyed with the way the appmod path has changed over
> releases - I'm not blaming - it's probably my fault .... maybe.
> 
> Anyway - this is precisely the sort of thing a decent tests suite
> would cover. Any takers !!!!!

Actually I'm pretty happy about the current behaviour. If
the appmod path is "/" then appmoddata will contain the leading
slash otherwise not. Makes sense and IMHO is correct.

Thus I'd like to keep as is.


Cheers


/klacke

On 2/5/07, Claes Wikstrom <klacke@...> wrote:
> Claes Wikstrom wrote:
>
> > Ouch - not ok. Will fix immediately. I'm actually getting
> > pretty annoyed with the way the appmod path has changed over
> > releases - I'm not blaming - it's probably my fault .... maybe.
> >
> > Anyway - this is precisely the sort of thing a decent tests suite
> > would cover. Any takers !!!!!
>
> Actually I'm pretty happy about the current behaviour. If
> the appmod path is "/" then appmoddata will contain the leading
> slash otherwise not. Makes sense and IMHO is correct.
>
> Thus I'd like to keep as is.

That is fine with me, I agree that it makes sense. I was just
mentioning the difference between versions mostly for other people's
benefit, I had already worked around it before I mentioned it.

Personally all of my code that pokes into Yaws records goes through a
single module in my application, so it's easy to have functions that
will accept either way that Yaws behaves.

-bob