Rounding Floats : Beware

General erlang point, rather than CB-specific, but . . .

I made a casual google search on erlang rounding and turned up the following:

  "round/1 returns the closest integer to x, rounding to even when x is halfway between two integers."

(TrapExit http://www.trapexit.org/Floating_Point_Rounding - though the error is repeated in several other places)

NOT TRUE ( well not on my system anyway)

ROUND HALF TO EVEN is a very common standard practice, particularly in bookkeeping applications (and according to Wikipedia, "This is the default rounding mode used in IEEE 754 computing functions and operators").

However, erlang uses ROUND HALF AWAY FROM ZERO.

From my shell:

(eg <at> testhost)12> round (3.5).
4
(eg <at> testhost)13> round (4.5).
5
(eg <at> testhost)14> round(-3.5).
-4
(eg <at> testhost)15> round(-4.5).
-5
(eg <at> testhost)16>

This may catch some by surprise - it certainly did me. It happens that for this application neither is what I need (I want ROUND HALF UP, but that is a different story).

The official documentation is characteristically coy on the subject, not commenting on the strategy used. But how hard is it to start up an erlang shell and try it for goodness sake?

Anyway - a word to the wise.     Dont believe all you read

graeme

PS
(With my generous hat on . . . Perhaps the behaviour was different in the past.  )

--
You received this message because you are subscribed to the Google Groups "ChicagoBoss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chicagoboss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Turn off sessions for controller(s)

"class methods" for models

Hi,

I'd like to do something like

student:some_utility_function(Foo, Bar) without it being fiddled with
in the parse transform stuff.

Ideas?

--
David N. Welton

http://www.welton.it/davidw/

http://www.dedasys.com/

--

-- 
You received this message because you are subscribed to the Google Groups "ChicagoBoss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chicagoboss+unsubscribe@...
For more options, visit https://groups.google.com/groups/opt_out.

start gen_server module

Enum sessions

Creating a page within a page

Creating multiple pages associated with a controller.

redefined function error

Here's a weird one...

Model 'student':

-has({exams, many}).

first_exam() ->
    lists:nth(1, completed_exams()).

That doesn't complain on my computer, but on my colleague's computer,
it does.  Same erlang version, same code.  Actually, to be more
precise, it does not compile for him, but he made a new account for me
on the computer, and it compiles fine.

==> myapp_web (pre_compile)
ERROR: pre_compile failed while processing
/home/albert64/projects64/myapp/myapp_web:
{'EXIT',{{badmatch,{error,[{"/home/albert64/projects64/myapp/myapp_web/src/model/student.erl",
                            [{0,erl_lint,{redefine_function,{first_exam,1}}},
                             {0,erl_lint,
                              {redefine_function,{last_exam,1}}}]}]}},
         [{boss_load,load_all_modules,3,
                     [{file,"src/boss/boss_load.erl"},{line,30}]},
          {boss_load,load_all_modules_and_emit_app_file,2,
                     [{file,"src/boss/boss_load.erl"},{line,44}]},
          {boss_rebar,compile,4,

[{file,"/home/albert64/projects64/ChicagoBoss/priv/rebar/boss_rebar.erl"},
                       {line,85}]},
          {boss_plugin,pre_compile,2,
                       [{file,"priv/rebar/boss_plugin.erl"},{line,105}]},
          {rebar_core,run_modules,4,[]},
          {rebar_core,execute,5,[]},
          {rebar_core,process_dir1,6,[]},
          {rebar_core,process_commands,2,[]}]}}

I think it's good that it complains, but I'm struggling to figure out
what might be different in his environment that makes it do this.

Any ideas?

--
David N. Welton

http://www.welton.it/davidw/

http://www.dedasys.com/

--

-- 
You received this message because you are subscribed to the Google Groups "ChicagoBoss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chicagoboss+unsubscribe@...
For more options, visit https://groups.google.com/groups/opt_out.

eunit tests won't start with with additional applications

More testing woes..

I have a couple of extra applications being managed by CB (inets and
bcrypt), they are included in src/<appname>.app.src and they have
their own modules under priv/init, namely <appname>_02_inets.erl and
<appname>_03_bcrypt.erl.

As long as those extra applications are included in <appname>.app.src,
eunit tests won't start (rebar and priv/rebar/boss_plugin.erl are from
0.8.4, and the result is the same whether I use CB 0.8.4 or 0.8.6). By
printing out log messages, I get lots and lots of

INFO:  Boss App <appname> still not started, waiting...

so boss_rebar:boss_start_wait() loops forever, as <appname> seemingly
isn't started. If I omit inets and bcrypt from <appname>.app.src,
<appname> starts, but running the tests fails due to missing
dependencies.

The applications list in <appname>.app.src reads

 {applications, [kernel, stdlib, crypto, boss, inets, bcrypt]}

(I've also tried omitting boss here, but the result stays the same)
and the init functions of the inets and bcrypt init modules read,
respectively

init() ->
    inets:start(),
    {ok, []}.

and

init() ->
    bcrypt:start(),
    {ok, []}.

I have had the tests running before, but I don't recall whether this
was before or after I added the extra applications, and I've regularly
updated CB to the latest versions.

Does anybody have any idea what's going on here? Should I start the
applications differently or do I need something extra in
boss.test.config? Or are eunit tests with extra applications broken in
(at least) 0.8.4?

Thanks,

- Fredrik

--

-- 
You received this message because you are subscribed to the Google Groups "ChicagoBoss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chicagoboss+unsubscribe@...
For more options, visit https://groups.google.com/groups/opt_out.

eunit testing broken in 0.8.6?

I can't seem to get eunit tests to run in 0.8.6. With rebar and
priv/rebar/boss_plugin.erl from 0.8.4, they do. Is eunit testing
broken in 0.8.6 or am I doing something wrong?

Freshly set-up CB 0.8.6, Erlang R16B, wholly new app, foo. From what i
can see there seems to be some disagreement regarding rebar's API:

$ ./rebar boss c=test_eunit
==> foo (boss)
ERROR: boss failed while processing /home/fdv/src/tests/cb/foo:
{'EXIT',
 {undef,
  [{rebar_erlc_compiler,doterl_compile,
   ...

Can anybody shed some light on this?

Thanks,
- Fredrik

--

-- 
You received this message because you are subscribed to the Google Groups "ChicagoBoss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chicagoboss+unsubscribe@...
For more options, visit https://groups.google.com/groups/opt_out.

Using boss_migrate for initial data