[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
- Subject: Re: clumsiness of pcall() syntax and how we deal with it
- From: Daniel Silverstone <dsilvers@...>
- Date: Tue, 28 Apr 2015 11:32:40 +0100
On Tue, Apr 28, 2015 at 13:19:04 +0300, Konstantin Osipov wrote:
> The first problem is in ambiguity of the second argument.
> In case there is no error, it's the return value of the function.
> Otherwise, it's a string with a message.
I tend to use 'ok, result = pcall(xyz, blah)' and then:
if ok then
...use result as useful result...
else
...result is the error...
end
> How to name the second argument? And what if a function returns
> multiple arguments?
Multiple returns is a toughie and I will admit I tend not to use
them in cases where I want to return errors too. Then again I tend
to code functions which return an ok, result tuple most of the time
and then deal with wrappering them in assert() when I'd rather they
threw an error.
> The solution we adopted is to never treat the second argument
> as an error, but use a separate API to fetch the error.
This is a not-uncommon approach in use in many other languages. When
I did a database wrapper in another place and time I had something
equivalent to:
if not db_handle:insert(.....) then
do_something_with(db_handle:get_last_error())
end
> As can be seen from the snippet, we also extended the error
> handling logic to be able to raise and check additional
> attributes of the error, such as code or error number,
> and added an ability to re-throw an already raised error,
> without change.
Did you extend it to actually wrapper xpcall and record error location and
stack trace too? That might be of use to debuggers.
> It would be nice to see support for these needs in some future
> version of the Lua language. Especially since all the mechanics
> is already there, and we're only talking about making it
> easier to use.
There have been various attempts at using token filters to introduce more
python or C++ like exception handling into the language. None were widely
enough liked to warrant a language change though.
D.
--
Daniel Silverstone http://www.digital-scurf.org/
PGP mail accepted and encouraged. Key Id: 3CCE BABE 206C 3B69