www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - regressions dmd-113 -> dmd-118

reply Thomas Kuehne <thomas-dloop kuehne.thisisspam.cn> writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

XFAIL->ERROR
http://dstress.kuehne.cn/nocompile/const_04.d

XFAIL->XPASS
http://dstress.kuehne.cn/nocompile/main_06.d

FAIL->ERROR
http://dstress.kuehne.cn/run/html_htm_02.HTM

full listing: (1010K)
http://dstress.kuehne.cn/www/results.html

Thomas


-----BEGIN PGP SIGNATURE-----

iD8DBQFCOZdP3w+/yD4P9tIRAgUwAKClS9RDXk17KPioPLlctyK/TuZaBACfcReG
7XymsaxU1PdPmlIz2aHXDvc=
=rG9a
-----END PGP SIGNATURE-----
Mar 17 2005
parent reply Stewart Gordon <smjg_1998 yahoo.com> writes:
Thomas Kuehne wrote:

 -----BEGIN PGP SIGNED MESSAGE-----
 Hash: SHA1
 
 XFAIL->ERROR
 http://dstress.kuehne.cn/nocompile/const_04.d
 
 XFAIL->XPASS
 http://dstress.kuehne.cn/nocompile/main_06.d


That's a fix, not a regression.  It's in the change log for 0.116, and 
AIUI in the spec a while before this.

I've done an Excel file to show the trend in success/stability rates 
(hope you don't mind):

http://smjg.port5.com/pr/d/dstress.xls

Stewart.

-- 
My e-mail is valid but not my primary mailbox.  Please keep replies on 
the 'group where everyone may benefit.
Mar 17 2005
parent reply Thomas Kuehne <thomas-dloop kuehne.thisisspam.cn> writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stewart Gordon schrieb am Thu, 17 Mar 2005 16:33:53 +0000:

 Thomas Kuehne wrote:



 XFAIL->XPASS
 http://dstress.kuehne.cn/nocompile/main_06.d





 That's a fix, not a regression.  It's in the change log for 0.116, and 
 AIUI in the spec a while before this.












The code for b is correct but a's code is illegal.

http://digitalmars.com/d/statement.html




 I've done an Excel file to show the trend in success/stability rates 
 (hope you don't mind):

 http://smjg.port5.com/pr/d/dstress.xls


Great!

What is the meaning of the colors on the "Result" sheet?

Is this an OpenOffice bug or is the 0.118 version label missing on the
"DMD Trend" sheet?

Thomas



-----BEGIN PGP SIGNATURE-----

iD8DBQFCObnH3w+/yD4P9tIRAkMKAJ0U7W4jMOAQ45iKQlcOhxQ9VUTq+ACeLgjg
oyPVeaex2RYImyQTrHvCOYk=
=VKpt
-----END PGP SIGNATURE-----
Mar 17 2005
parent reply Stewart Gordon <smjg_1998 yahoo.com> writes:
Thomas Kuehne wrote:

 Stewart Gordon schrieb am Thu, 17 Mar 2005 16:33:53 +0000:
 

 Thomas Kuehne wrote:


 

 XFAIL->XPASS http://dstress.kuehne.cn/nocompile/main_06.d




 

 That's a fix, not a regression.  It's in the change log for 0.116,
 and AIUI in the spec a while before this.


 



That shouldn't compile - it has no return statement.



 
 The code for b is correct but a's code is illegal.


http://www.digitalmars.com/d/statement.html#return

"Expression is allowed even if the function specifies a void return
type. The Expression will be evaluated, but nothing will be returned."






A return statement isn't an expression.  But good question ... _should_
they be illegal as return statements as well?


 I've done an Excel file to show the trend in success/stability
 rates (hope you don't mind):
 
 http://smjg.port5.com/pr/d/dstress.xls


 
 Great!
 
 What is the meaning of the colors on the "Result" sheet?


Red - worse than the immediately previous version
Plum - worse than the best prior version

This was done using the Conditional Formatting feature of Excel.  Notice 
that a lot of the reds are where it hops between DMD and GDC
versions.


 Is this an OpenOffice bug or is the 0.118 version label missing on
 the "DMD Trend" sheet?


It looks fine to me.  Must be a bug in OpenOffice or its Excel filter.
Does OpenOffice show the labels inclined at 45° as they're supposed to
be?  And does it show all the 18 other version labels?

Stewart.

-- 
My e-mail is valid but not my primary mailbox.  Please keep replies on
the 'group where everyone may benefit.
Mar 17 2005
next sibling parent Nick <Nick_member pathlink.com> writes:
In article <d1ches$1h10$1 digitaldaemon.com>, Stewart Gordon says...

http://www.digitalmars.com/d/statement.html#return

"Expression is allowed even if the function specifies a void return
type. The Expression will be evaluated, but nothing will be returned."


What!? I thought that "feature" was removed long ago. Is there any reason why
this is allowed? Seems useless and strange to me.

Nick
Mar 17 2005
prev sibling next sibling parent reply "Unknown W. Brackets" <unknown simplemachines.org> writes:
That's only true when warnings are enabled, isn't it?

-[Unknown]



 That shouldn't compile - it has no return statement.
Mar 17 2005
parent Stewart Gordon <smjg_1998 yahoo.com> writes:
Unknown W. Brackets wrote:

 That's only true when warnings are enabled, isn't it?


<snip top of upside-down reply>

Every function with a non-void return type is required to have at least 
one return statement.  The warnings simply tell you if not every path 
through the function has a return statement.

For example, this is illegal

     int qwert(int yuiop) {}

whereas this sparks a warning

     int qwert(int yuiop) {
         if (yuiop > 0) return yuiop;
     }

and this would compile without warning or error

     int qwert(int yuiop) {
         if (yuiop > 0) return yuiop;
         return 0;
     }

Stewart.

-- 
My e-mail is valid but not my primary mailbox.  Please keep replies on 
the 'group where everyone may benefit.
Mar 18 2005
prev sibling parent reply Thomas Kuehne <thomas-dloop kuehne.thisisspam.cn> writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stewart Gordon schrieb am Thu, 17 Mar 2005 18:14:51 +0000:

 Thomas Kuehne wrote:

 Stewart Gordon schrieb am Thu, 17 Mar 2005 16:33:53 +0000:
 

 Thomas Kuehne wrote:


 

 XFAIL->XPASS http://dstress.kuehne.cn/nocompile/main_06.d




 

 That's a fix, not a regression.  It's in the change log for 0.116,
 and AIUI in the spec a while before this.


 



 That shouldn't compile - it has no return statement.


....



 
 The code for b is correct but a's code is illegal.


 http://www.digitalmars.com/d/statement.html#return

 "Expression is allowed even if the function specifies a void return
 type. The Expression will be evaluated, but nothing will be returned."






 A return statement isn't an expression.  But good question ... _should_
 they be illegal as return statements as well?


"return;" and "return something();" are statements but
the "something()" of "return something();" is an expression.


 I've done an Excel file to show the trend in success/stability
 rates (hope you don't mind):
 
 http://smjg.port5.com/pr/d/dstress.xls







 Is this an OpenOffice bug or is the 0.118 version label missing on
 the "DMD Trend" sheet?


 It looks fine to me.  Must be a bug in OpenOffice or its Excel filter.
 Does OpenOffice show the labels inclined at 45° as they're supposed to
 be?  And does it show all the 18 other version labels?


Everything else is fine.

Thomas

-----BEGIN PGP SIGNATURE-----

iD8DBQFCOmvx3w+/yD4P9tIRAkiTAKDGdkVAUbY+AO7YwYwziUVFkREktACgszOu
t6hZJchZgXlqzxGwd/AaEBM=
=dq6c
-----END PGP SIGNATURE-----
Mar 17 2005
parent reply Stewart Gordon <smjg_1998 yahoo.com> writes:
Thomas Kuehne wrote:

 Stewart Gordon schrieb am Thu, 17 Mar 2005 18:14:51 +0000:
 

 Thomas Kuehne wrote:




<snip>

 http://digitalmars.com/d/statement.html




 
 A return statement isn't an expression.  But good question ... 
 _should_ they be illegal as return statements as well?


 
 "return;" and "return something();" are statements but
 the "something()" of "return something();" is an expression.


<snip>

I meant to say:
A return statement isn't an expression statement.

ExpressionStatement:
     Expression ;

ReturnStatement:
     return ;
     return Expression ;

The paragraph you quoted from the spec is talking about the former.

Stewart.

-- 
My e-mail is valid but not my primary mailbox.  Please keep replies on 
the 'group where everyone may benefit.
Mar 18 2005
parent Thomas Kuehne <thomas-dloop kuehne.thisisspam.cn> writes:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stewart Gordon schrieb am Fri, 18 Mar 2005 10:47:51 +0000:

 Thomas Kuehne wrote:

 Stewart Gordon schrieb am Thu, 17 Mar 2005 18:14:51 +0000:
 

 Thomas Kuehne wrote:







 http://digitalmars.com/d/statement.html




 
 A return statement isn't an expression.  But good question ... 
 _should_ they be illegal as return statements as well?


 
 "return;" and "return something();" are statements but
 the "something()" of "return something();" is an expression.





 I meant to say:
 A return statement isn't an expression statement.


For normal returns there is no problem with no-effect expressions,
but for void-returns these expressions should be illegal.

I'd rather opt for removing void-return than to start complicating the
rules.

Thomas


-----BEGIN PGP SIGNATURE-----

iD8DBQFCOtbR3w+/yD4P9tIRAoFaAKDMEV+DK5TdrjIWsPXGNKkel3T4hwCfWEil
sP2dZtre8wC0yqmW+MQ/zHk=
=HYa8
-----END PGP SIGNATURE-----
Mar 18 2005