www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - version(CONTRACTS) for contracts?

reply Bill Baxter <dnewsgroup billbaxter.com> writes:
Is there a version statement that corresponds to whether or not 
contracts are enabled or not?

I'd like to do something like

   in { _check_data_structure_integrity() }

for a bunch of my methods in my class, then something like

    version(Contracts) {
       _check_data_structure_integrity() {
          // do a bunch of complicated stuff here
          // and assert if anything's messed up.
       }
    }

--bb
Nov 08 2006
parent reply BCS <fizzle pathlink.com> writes:
I think what you are trying for is what class invariants do.

http://www.digitalmars.com/d/class.html#invariants

== Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?
 I'd like to do something like
    in { _check_data_structure_integrity() }
 for a bunch of my methods in my class, then something like
     version(Contracts) {
        _check_data_structure_integrity() {
           // do a bunch of complicated stuff here
           // and assert if anything's messed up.
        }
     }
 --bb
Nov 08 2006
parent reply Bill Baxter <dnewsgroup billbaxter.com> writes:
BCS wrote:

 I think what you are trying for is what class invariants do.
 
 http://www.digitalmars.com/d/class.html#invariants
 
 == Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?
 I'd like to do something like
    in { _check_data_structure_integrity() }
 for a bunch of my methods in my class, then something like
     version(Contracts) {
        _check_data_structure_integrity() {
           // do a bunch of complicated stuff here
           // and assert if anything's messed up.
        }
     }
 --bb


 


Class invariants. Indeed so!

Still I don't think that covers all use cases.  If my invariant is very 
slow to compute and I have some methods that are very frequently called, 
then I'm not going to want to check it for every single method call.  It 
could make debugging way too slow.  Plus, if I know these three methods 
are the only three that affect this invariant then only those three 
methods should trigger an invariant check.

So I think something more fine grained than class invariants is needed. 
  Something like version(_DContracts) would do it.

--bb
Nov 08 2006
next sibling parent Daniel Keep <daniel.keep.lists gmail.com> writes:
Bill Baxter wrote:

 BCS wrote:

 I think what you are trying for is what class invariants do.

 http://www.digitalmars.com/d/class.html#invariants

 == Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?
 I'd like to do something like
    in { _check_data_structure_integrity() }
 for a bunch of my methods in my class, then something like
     version(Contracts) {
        _check_data_structure_integrity() {
           // do a bunch of complicated stuff here
           // and assert if anything's messed up.
        }
     }
 --bb




 
 Class invariants. Indeed so!
 
 Still I don't think that covers all use cases.  If my invariant is very
 slow to compute and I have some methods that are very frequently called,
 then I'm not going to want to check it for every single method call.  It
 could make debugging way too slow.  Plus, if I know these three methods
 are the only three that affect this invariant then only those three
 methods should trigger an invariant check.
 
 So I think something more fine grained than class invariants is needed.
  Something like version(_DContracts) would do it.
 
 --bb


Why don't you do something like this?


 in { version(SlowInvariants) _check_stuff(); }

 // ...

 version(SlowInvariants) {
     void _check_stuff() {
         // ...
     }
 }


Then, when compiling, stick a -version=SlowInvariants on the command
line.  You should be able to set up a release and debug target in bud so
you only need to do this once.

	-- Daniel

-- 
Unlike Knuth, I have neither proven or tried the above; it may not even
make sense.

v2sw5+8Yhw5ln4+5pr6OFPma8u6+7Lw4Tm6+7l6+7D
i28a2Xs3MSr2e4/6+7t4TNSMb6HTOp5en5g6RAHCP  http://hackerkey.com/
Nov 08 2006
prev sibling next sibling parent reply Derek Parnell <derek nomail.afraid.org> writes:
On Thu, 09 Nov 2006 12:06:34 +0900, Bill Baxter wrote:


 BCS wrote:

 I think what you are trying for is what class invariants do.
 
 http://www.digitalmars.com/d/class.html#invariants
 
 == Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?
 I'd like to do something like
    in { _check_data_structure_integrity() }
 for a bunch of my methods in my class, then something like
     version(Contracts) {
        _check_data_structure_integrity() {
           // do a bunch of complicated stuff here
           // and assert if anything's messed up.
        }
     }
 --bb


 


 
 Class invariants. Indeed so!
 
 Still I don't think that covers all use cases.  If my invariant is very 
 slow to compute and I have some methods that are very frequently called, 
 then I'm not going to want to check it for every single method call.  It 
 could make debugging way too slow.  Plus, if I know these three methods 
 are the only three that affect this invariant then only those three 
 methods should trigger an invariant check.
 
 So I think something more fine grained than class invariants is needed. 
   Something like version(_DContracts) would do it.
 
 --bb


I can do this with the next Bud. 

If "-release" is used, I can add "version=_Drelease" to the command line.
If "-release" is *not* used, I can add "version=_Dcontracts" to the command
line.
If "-unittest" is used, I can add "version=_Dunittest" to the command line.
If "-O" is used, I can add "version=_Doptimized" to the command line.

Is anyone interested in this idea? Does it need tweaking? 

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
"Down with mediocrity!"
9/11/2006 2:13:20 PM
Nov 08 2006
next sibling parent reply Bill Baxter <dnewsgroup billbaxter.com> writes:
Derek Parnell wrote:

 On Thu, 09 Nov 2006 12:06:34 +0900, Bill Baxter wrote:
 



 So I think something more fine grained than class invariants is needed. 
   Something like version(_DContracts) would do it.

 --bb


 
 I can do this with the next Bud. 
 
 If "-release" is used, I can add "version=_Drelease" to the command line.
 If "-release" is *not* used, I can add "version=_Dcontracts" to the command
 line.
 If "-unittest" is used, I can add "version=_Dunittest" to the command line.
 If "-O" is used, I can add "version=_Doptimized" to the command line.
 
 Is anyone interested in this idea? Does it need tweaking? 
 


Sounds sweet to me.

Not sure if _Doptimized is really useful though.  But if that makes 
sense, then probably something should be done for -g too.
And what about -run?  I can see a use for that more readily than I can 
for -O or -g.

I thought I remembered posting something about a wish for a version to 
go with unittests and somebody posted some kind of patch right away.  I 
couldn't find it in the archives though.  Oh yes... it was a bug not a 
ng post.  That's why I couldn't find it:  here it is:

   http://d.puremagic.com/issues/show_bug.cgi?id=458

--bb
Nov 08 2006
parent Derek Parnell <derek nomail.afraid.org> writes:
On Thu, 09 Nov 2006 12:52:03 +0900, Bill Baxter wrote:

Ok, a more generic technique might be to have Bud add "version=D_<x>" for
each simple compiler switch supplied. Such that if "-run" was supplied Bud
would add "version=D_run", etc...

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
"Down with mediocrity!"
9/11/2006 2:59:03 PM
Nov 08 2006
prev sibling parent reply Hasan Aljudy <hasan.aljudy gmail.com> writes:
Derek Parnell wrote:

 On Thu, 09 Nov 2006 12:06:34 +0900, Bill Baxter wrote:
 

 BCS wrote:

 I think what you are trying for is what class invariants do.

 http://www.digitalmars.com/d/class.html#invariants

 == Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?
 I'd like to do something like
    in { _check_data_structure_integrity() }
 for a bunch of my methods in my class, then something like
     version(Contracts) {
        _check_data_structure_integrity() {
           // do a bunch of complicated stuff here
           // and assert if anything's messed up.
        }
     }
 --bb




 Class invariants. Indeed so!

 Still I don't think that covers all use cases.  If my invariant is very 
 slow to compute and I have some methods that are very frequently called, 
 then I'm not going to want to check it for every single method call.  It 
 could make debugging way too slow.  Plus, if I know these three methods 
 are the only three that affect this invariant then only those three 
 methods should trigger an invariant check.

 So I think something more fine grained than class invariants is needed. 
   Something like version(_DContracts) would do it.

 --bb


 
 I can do this with the next Bud. 
 
 If "-release" is used, I can add "version=_Drelease" to the command line.
 If "-release" is *not* used, I can add "version=_Dcontracts" to the command
 line.
 If "-unittest" is used, I can add "version=_Dunittest" to the command line.
 If "-O" is used, I can add "version=_Doptimized" to the command line.
 
 Is anyone interested in this idea? Does it need tweaking? 
 


But why? I don't get it.
If -release is passed, then contracts won't run anyway, so who needs a 
version identifier for it?
Nov 08 2006
parent Bill Baxter <dnewsgroup billbaxter.com> writes:
 But why? I don't get it.
 If -release is passed, then contracts won't run anyway, so who needs a 
 version identifier for it?



You must have skipped that part of the discussion.
Because I've got some code that I want to share between different 
contracts, but it should't be compiled in when contracts aren't being used.

--bb
Nov 08 2006
prev sibling parent Stewart Gordon <smjg_1998 yahoo.com> writes:
Bill Baxter wrote:

 BCS wrote:

 I think what you are trying for is what class invariants do.

 http://www.digitalmars.com/d/class.html#invariants

 == Quote from Bill Baxter (dnewsgroup billbaxter.com)'s article

 Is there a version statement that corresponds to whether or not
 contracts are enabled or not?






<snip>

 Still I don't think that covers all use cases.  If my invariant is very 
 slow to compute and I have some methods that are very frequently called, 
 then I'm not going to want to check it for every single method call.  It 
 could make debugging way too slow.  Plus, if I know these three methods 
 are the only three that affect this invariant then only those three 
 methods should trigger an invariant check.


So _check_data_structure_integrity is compiled only if DBC is enabled? 
That's really a matter of compiler/linker optimisation.  A design 
principle of D is that language features to compensate for compiler 
naivety are, in general, omitted.


 So I think something more fine grained than class invariants is needed. 
  Something like version(_DContracts) would do it.


Why the name _DContracts?


-- 
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GCS/M d- s:-  C++  a->--- UB  P+ L E  W++  N+++ o K-  w++  O? M V? PS- 
PE- Y? PGP- t- 5? X? R b DI? D G e++++ h-- r-- !y
------END GEEK CODE BLOCK------

My e-mail is valid but not my primary mailbox.  Please keep replies on 
the 'group where everyone may benefit.
Nov 12 2006