Bug with variadic parameters

[Night-walker]

(dao) load net.http
= none
(dao) routine f(...: tuple<enum,int|string|list<string>> as args){
..... http.initRequest('GET', 'some', args, ...)
..... }
[[ERROR]] in file "interactive codes":
  At line 1 : Invalid function definition --- " f() ";
  At line 2 : Invalid virtual machine instruction --- " MCALL:34,17412,39 ";
In code snippet:
      6 :  MOVE_PP     :    32 ,     0 ,    38 ;     2;   args
>>    7 :  MCALL       :    34 , 17412 ,    39 ;     2;   http. initRe...('GET', '...'...
      8 :  RETURN      :    39 ,     1 ,     0 ;     2;   http. initRequest('GET',...
  ** Invalid parameter type --- " '...:tuple<enum,int|string|list<string>>' for 'tuple<enum,int|string|list<string>>' ";
     Assuming  : routine initRequest(method:string,url:string,...:tuple<enum,int|string|list<string>>)=>string;
     Reference : file "http";

[daokoder]

Fixed.

[Night-walker]

(dao) load net.http
= none
(dao) routine request(method: string, url: string, ...: tuple<enum,int|string|list<string>> as fields){
..... http.initRequest('GET', url, fields, ...)
..... }
[[ERROR]] in file "interactive codes":
  At line 1 : Invalid function definition --- " request() ";
  At line 2 : Invalid virtual machine instruction --- " MCALL:34,17412,39 ";
In code snippet:
      6 :  MOVE_PP     :    32 ,     0 ,    38 ;     2;   fields
>>    7 :  MCALL       :    34 , 17412 ,    39 ;     2;   http. initRe...('GET', u...
      8 :  RETURN      :    39 ,     1 ,     0 ;     2;   http. initRequest('GET',...
  ** Invalid parameter type --- " 'string' for 'tuple<enum,int|string|list<string>>' ";
     Assuming  : routine initRequest(method:string,url:string,...:tuple<enum,int|string|list<string>>)=>string;
     Reference : file "http";

[daokoder]

fields also includes the method and url parameters, so they are also expanded there, and their types are not correct for the variadic parameter of http.initRequest().

[Night-walker]

I would expect fields to capture only the last, variadic part of routine parameters, as there is arguably no use in capturing other, fixed arguments.

[daokoder]

The choice was made to simplify the implementation, otherwise, we will need to add a dedicated virtual instruction or modify the DVM_TUPLE instruction for this. But if this is too confusing, I will change it.

[Night-walker]

I could never thought it worked that way. Either the semantics or the syntax should better be changed.

[dumblob]

The choice was made to simplify the implementation, otherwise, we will need to add a dedicated virtual instruction or modify the DVM_TUPLE instruction for this. But if this is too confusing, I will change it.

I came across this and have to admit it's really confusing. Any plans to change it?