RE: [sv-ec] RE: Mantis 2358 - isssue 168

After review, I feel that my issues with 2358 have largely been addressed.
However, I still don't like the wording of the 2358 proposal:

"It shall be an error for a derived class to specify the arguments to the base class constructor both when the class is extended and with a super.new(...) call in the constructor." 

The wording is ambiguous in the case where the arguments are specified when the class is extended, whether it is illegal to call super.new at all, or it is illegal to call super.new with arguments. One might interpret this text to say that "super.new;" without arguments is allowed. 

Assuming that the intent was to disallow a call to super.new at all in this case, I would reword as follows:

"If the arguments are specified at the time the class is extended, it shall be an error to have an explicit super.new call in the subclass constructor. In this case, the compiler shall insert a call to super.new automatically, as specified in 8.14."

This still leaves open some of the more obscure issues brought up in 2358, such as regarding evaluation of the arguments, but I think those are outside the scope of this ballot comment, and should be split off into a separate Mantis item.

I am also OK with the proposal for 2274.

Thanks,
Shalom

> -----Original Message-----
> From: Neil.Korpusik@Sun.COM [mailto:Neil.Korpusik@Sun.COM] 
> Sent: Tuesday, April 28, 2009 9:30 PM
> To: Arturo Salz
> Cc: Rich, Dave; Mark Hartoog; Bresticker, Shalom; sv-ec@eda.org
> Subject: Re: [sv-ec] RE: Mantis 2358 - isssue 168
> 
> Hi Arturo,
> 
> Gord posted that clarification in the proposal for Mantis 2724.
> 
> From:
>     When using the super within new, super.new shall be the 
> first statement executed in the constructor.
> To:
>     A super.new call shall be the first statement executed in 
> the constructor.
> 
> 
> Neil
> 
> 
> 
> On 04/27/09 23:23, Arturo Salz wrote:
> > Dave,
> > 
> >  
> > 
> > We agree that the language is awkward and can be improved 
> (no point in 
> > beating a dead horseJ) but the intent I believe is clear - 
> Wasn't Gord 
> > going to suggest some improved verbiage.
> > 
> >  
> > 
> >             Arturo
> > 
> >  
> > 
> > *From:* owner-sv-ec@eda.org [mailto:owner-sv-ec@eda.org] 
> *On Behalf Of 
> > *Rich, Dave
> > *Sent:* Monday, April 27, 2009 7:13 PM
> > *To:* Mark Hartoog; Bresticker, Shalom; sv-ec@eda.org
> > *Subject:* RE: [sv-ec] RE: Mantis 2358 - isssue 168
> > 
> >  
> > 
> > Mark,
> > 
> >  
> > 
> > What does "using the super"  mean?
> > 
> >  
> > 
> > 
> --------------------------------------------------------------
> ----------
> > 
> > *From:* owner-sv-ec@server.eda.org 
> [mailto:owner-sv-ec@server.eda.org] 
> > *On Behalf Of *Mark Hartoog
> > *Sent:* Monday, April 27, 2009 4:08 PM
> > *To:* Bresticker, Shalom; Mark Hartoog; sv-ec@eda.org
> > *Subject:* [sv-ec] RE: Mantis 2358 - isssue 168
> > 
> >  
> > 
> > Shalom,
> > 
> >  
> > 
> > Section 8.14 says:
> > 
> >  
> > 
> > When using the *super *within *new*, *super*.*new *shall be 
> the first 
> > statement executed in the constructor. This
> > 
> > is because the superclass shall be initialized before the 
> current class 
> > and, if the user code does not provide an
> > 
> > initialization, the compiler shall insert a call to *super*.*new 
> > *automatically.
> > 
> >  
> > 
> > I don't see what is unclear about this.  It says that if the user 
> > doesn't have a call to super.new, it is automatically called.
> > 
> >  
> > 
> > If the super class constructor requires arguments, obviously the 
> > compiler cannot insert a call automatically. Section 8.16 says:
> > 
> >  
> > 
> > If the initialization method of the superclass requires 
> arguments, there 
> > are two choices: to always supply the
> > 
> > same arguments or to use the *super *keyword.
> > 
> >  
> > 
> > I guess this does not say it is illegal not to do one of 
> these choices.  
> > I could see adding a sentence to say when the super class 
> constructor 
> > requires arguments, it is illegal to not specify them in 
> one of the two 
> > ways.
> > 
> >  
> > 
> > *From:* Bresticker, Shalom [mailto:shalom.bresticker@intel.com]
> > *Sent:* Sunday, April 26, 2009 11:33 PM
> > *To:* Mark Hartoog; sv-ec@eda.org
> > *Subject:* RE: Mantis 2358 - isssue 168
> > 
> >  
> > 
> > Mark,
> > 
> >  
> > 
> > 2358 brings up a number of points that need clarification. 
> This is just 
> > one of them, and not even the point that was the main point of the 
> > ballot comment.
> > 
> >  
> > 
> > I would change the wording of your proposal, but that is a 
> secondary 
> > issue. It still leaves ambiguous the following points from 
> the Mantis, 
> > for example:
> > 
> >  
> > 
> > LRM is unclear where a user-defined sub-class constructor 
> automatically 
> > calls super.new() if it does not contain such a call 
> explicitly. The 
> > answer is yes, but a statement such as "The default 
> constructor first 
> > calls its base class constructor" can be interpreted as 
> saying that it 
> > is true only for default sub-class constructors. In fact, 
> one company 
> > published an incorrect interpretation.
> > 
> > What happens if super.new() requires arguments, but they are not 
> > specified in the extends statement, and the sub-class new() 
> implicitly 
> > calls super.new()? (Probably a compilation error.)
> > 
> > If the extends statement specifies the super.new arguments, 
> e.g., class 
> > EtherPacket extends Packet(5);, may a sub-class 
> user-defined constructor 
> > contain an explicit call to super.new()? If so, what is the 
> syntax, with 
> > or without arguments, super.new() or super.new(5)?
> > 
> > (Your proposal says that the last option would not be 
> legal, but still 
> > leaves ambiguous whether super.new is implicitly called in a 
> > user-defined constructor, whether an explicit call to super.new is 
> > legal, and what is the syntax if so.)
> > 
> >  
> > 
> > Regards,
> > 
> > Shalom
> > 
> > Shalom Bresticker
> > Intel LAD DA
> > Jerusalem, Israel
> > +972  2 589 6582 (office)
> > +972 54 721 1033 (cell)
> > 
> >  
> > 
> >  
> > 
> > 
> --------------------------------------------------------------
> ----------
> > 
> > *From:* owner-sv-ec@server.eda.org 
> [mailto:owner-sv-ec@server.eda.org] 
> > *On Behalf Of *Mark Hartoog
> > *Sent:* Monday, April 27, 2009 4:18 AM
> > *To:* sv-ec@eda.org
> > *Subject:* [sv-ec] Mantis 2358 - isssue 168
> > 
> > I discussed this issue with Arturo. He feels that the situation in 
> > Mantis 2358 should be an error.
> > 
> >  
> > 
> > I have added a proposal to 2358 to make it an error. The 
> proposal is:
> > 
> >  
> > 
> > At the end of section 8.16, Chaining constructors, add the 
> following 
> > paragraph:
> > 
> >  
> > 
> > It shall be an error for a derived class to specify the 
> arguments to the 
> > base class constructor both when the class is extended and with a 
> > *super.new*(...) call in the constructor.
---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.