[erlang-patches] dialyzer: Report specs defining 'incorrect' lists as return types

Fredrik fredrik@REDACTED
Mon Dec 17 11:53:13 CET 2012


Thanks for the clarification Stavros!
I've already been in contact with the one responsible for the "unused 
mnemosyne code" patch, I will try to sort out the compile column numbers 
later today.

BR Fredrik Gustafsson
Erlang OTP Team
On 12/17/2012 11:49 AM, Stavros Aronis wrote:
> Hello!
>
> The error that you got ("dialyzer:format_warning failed on line 295 
> Reason: function_clause") comes from the unexpected structure of the 
> file/line annotation due to the column number patch.
>
> Tyrying to run all the tests on master-pu however, I also got a 
> different error ("small_SUITE:init_per_suite failed Reason: 
> try_clause"), as trying to include mnesia in Dialyzer's PLT exposes 
> the following error there:
> mnesia.erl:2739: Call to missing or unexported function 
> mnesia:get_activity_id/0
>
> The missing function was removed by the unused mnemosyne code patch.
>
> Hope this helps!
>
> Stavros
>
>
>
>
> On Mon, Dec 17, 2012 at 11:42 AM, Fredrik <fredrik@REDACTED 
> <mailto:fredrik@REDACTED>> wrote:
>
>     Hello again,
>     So what you are saying is that it is the patch that is reporting
>     column numbers and the patch that is removing unused mnemosyne code?
>     Neither of these patches has been graduated, they are in
>     master-pu, the same as your patch.
>     I will discuss with the responsible developers of the applications
>     concerned here and get back to you.
>
>
>     BR Fredrik Gustafsson
>     Erlang OTP Team
>     On 12/17/2012 11:21 AM, Stavros Aronis wrote:
>>     Btw,
>>
>>     the different error I get on master-pu comes from a very weird
>>     error on mnesia:
>>
>>     On mnesia.erl, line 2739 there is a call to missing function
>>     mnesia:get_activity_id/0.
>>
>>     This function was removed in commit
>>     6f4c3d127e33d59ac87a3c3b600f7a7280c9674a, part of the patch that
>>     removes unused mnemosyne code.
>>
>>     Regards,
>>
>>
>>     On Mon, Dec 17, 2012 at 11:09 AM, Stavros Aronis
>>     <aronisstav@REDACTED <mailto:aronisstav@REDACTED>> wrote:
>>
>>         Hi Fredrik,
>>
>>         Running the tests on my branch (which is rebased on maint)
>>         does not yield any errors.
>>
>>         Running the tests on master-pu, I get a different error.
>>
>>         By manually running the test I can see that the error that
>>         you report comes from what I would guess is the new extension
>>         that reports column numbers along file lines.
>>
>>         Has this patch graduated on master? What is the suggested
>>         course of action?
>>
>>         Regards,
>>
>>         Stavros
>>
>>
>>         On Mon, Dec 17, 2012 at 10:22 AM, Fredrik <fredrik@REDACTED
>>         <mailto:fredrik@REDACTED>> wrote:
>>
>>             Hello Stavros!
>>             This fails:
>>             'small_SUITE : empty_list_infimum'
>>             with reason:
>>             dialyzer:format_warning failed on line 295 Reason:
>>             function_clause
>>             On all platforms..
>>             Please fix this, and I will re-fetch.
>>
>>
>>             BR Fredrik Gustafsson
>>             Erlang OTP Team
>>             On 12/14/2012 10:11 AM, Stavros Aronis wrote:
>>>             This patch enables Dialyzer to detect cases where a
>>>             function has a specification for a return type
>>>             containing some list with elements different than those
>>>             Dialyzer's own algorithm can infer e.g. a function
>>>             specified to return [atom()] when actually [[atom()]] is
>>>             returned.
>>>
>>>             Previously Dialyzer remained silent under normal
>>>             operation, seeing that these two types have the empty
>>>             list as a common element, so the spec was not
>>>             'completely irrelevant'.
>>>
>>>             Using this patch one can find discrepancies in the specs
>>>             of stdlib's sys.erl and erl_scan.erl. Kostis and I are
>>>             in contact with the OTP team for the best way fix those
>>>             as well.
>>>
>>>               git fetch git://github.com/aronisstav/otp.git
>>>             <http://github.com/aronisstav/otp.git> dialyzer-list-spec
>>>
>>>             Regards,
>>>
>>>             Stavros
>>>
>>>             ---
>>>
>>>             The new warnings are:
>>>
>>>             sys.erl:303: Invalid type specification for function
>>>             sys:handle_debug/4. The success typing is
>>>             ([{_,_}],_,_,_) -> [{_,_}]
>>>             erl_scan.erl:204: Invalid type specification for
>>>             function erl_scan:token_info/2. The success typing is
>>>             (_,'category' | 'column' | 'length' | 'line' |
>>>             'location' | 'symbol' | 'text' |
>>>             maybe_improper_list('category' | 'column' | 'length' |
>>>             'line' | 'location' | 'symbol' | 'text','category' |
>>>             'column' | 'length' | 'line' | 'location' | 'symbol' |
>>>             'text' | [])) -> 'undefined' |
>>>             maybe_improper_list({_,_},'undefined' | [] | {_,_}) | {_,_}
>>>             erl_scan.erl:242: Invalid type specification for
>>>             function erl_scan:attributes_info/2. The success typing
>>>             is (_,'column' | 'length' | 'line' | 'location' | 'text'
>>>             | maybe_improper_list('column' | 'length' | 'line' |
>>>             'location' | 'text','column' | 'length' | 'line' |
>>>             'location' | 'text' | [])) -> 'undefined' |
>>>             maybe_improper_list({_,_},'undefined' | [] | {_,_}) | {_,_}
>>>
>>>
>>>
>>>
>>>             _______________________________________________
>>>             erlang-patches mailing list
>>>             erlang-patches@REDACTED  <mailto:erlang-patches@REDACTED>
>>>             http://erlang.org/mailman/listinfo/erlang-patches
>>
>>
>>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-patches/attachments/20121217/7d7ddcfc/attachment.htm>


More information about the erlang-patches mailing list