| Re: Bug in SWITCH - ELSE: statement? [message #56230] |
Fri, 12 October 2007 13:56 |
R.G.Stockwell
Messages: 163
Registered: October 2004
|
Senior Member |
|
|
"Matt" <savoie@nsidc.org> wrote in message
news:1192221602.770405.43150@q5g2000prf.googlegroups.com...
>
> JD,
>
> I think it's documentation bug. Look closely at the original poster's
> emphasis.
>
>>> **The ELSE statement is executed only if none of the preceding statement
>>> expressions match.**
>
> That line of documentation is inconsistent with the way switch works
> in IDL.
>
> Of course Jean H just pointed out.
>
> "SWITCH executes the first matching statement and any following
> statements in
> the SWITCH block. Once a match is found in the SWITCH block, execution
> falls
> through to any remaining statements. For this reason, the BREAK
> statement is
> commonly used within SWITCH statements to force an immediate exit from
> the
> SWITCH block."
>
> So the documentation is orthogonal as far as I can tell, and the first
> statement is the wrong one.
Right, looks like someone copy and pasted that paragraph from the
CASE statement help file.
It is correctly described in the previous paragraph though.
For the SWITCH help, it really should state:
**The ELSE statement is executed only if none of the preceding statement
expressions match,
or it any of the preceding statement expressions match.**
:)
Cheers,
bob
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56231 is a reply to message #56230] |
Fri, 12 October 2007 13:40  |
Matt[2]
Messages: 69
Registered: March 2007
|
Member |
|
|
JD,
I think it's documentation bug. Look closely at the original poster's
emphasis.
>> **The ELSE statement is executed only if none of the preceding statement expressions match.**
That line of documentation is inconsistent with the way switch works
in IDL.
Of course Jean H just pointed out.
"SWITCH executes the first matching statement and any following
statements in
the SWITCH block. Once a match is found in the SWITCH block, execution
falls
through to any remaining statements. For this reason, the BREAK
statement is
commonly used within SWITCH statements to force an immediate exit from
the
SWITCH block."
So the documentation is orthogonal as far as I can tell, and the first
statement is the wrong one.
Matt
--
Matthew Savoie - Scientific Programmer
National Snow and Ice Data Center
http://nsidc.org
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56232 is a reply to message #56231] |
Fri, 12 October 2007 12:43 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Thu, 11 Oct 2007 09:44:27 +0000, Spon wrote:
> Hi everyone,
>
> can someone explain this to me please?
>
>> [quoted text muted]
> The ELSE clause of the SWITCH statement is optional. If included, it
> matches any selector expression, causing its code to be executed. For
> this reason, it is usually written as the last clause in the switch
> statement. **The ELSE statement is executed only if none of the
> preceding statement expressions match.** If an ELSE clause is not
> included and none of the values match the selector, program execution
> continues immediately below the SWITCH without executing any of the
> SWITCH statements.
>
> But, for example:
>
> PRO SWITCHTEST
>
> N = FIX (6 * RANDOMU (S, 1) ) + 1
>
> SWITCH N OF
> 1: PRINT, 'One'
> 2: PRINT, 'Two or less'
> 3: PRINT, 'Three or less'
> 4: PRINT, 'Four or less'
> 5: PRINT, 'Five or less'
> 6: PRINT, 'Six or less'
> ELSE: PRINT, 'Are you using loaded dice?'
> ENDSWITCH
>
> PRINT, N
> END
You probably wanted CASE, which has an implicit BREAK after each
statement. Or put the BREAKs in yourself:
1: begin
print, 'One'
break
end
SWITCH is really useful only when you want to "fall through" to multiple
processing steps. If you do want to fall through, you must BREAK
somewhere before ELSE:
SWITCH N OF
1: PRINT, 'One'
2: PRINT, 'Two or less'
3: PRINT, 'Three or less'
4: PRINT, 'Four or less'
5: PRINT, 'Five or less'
6: begin
PRINT, 'Six or less'
break
end
ELSE: PRINT, 'Are you using loaded dice?'
ENDSWITCH
Yes, it's ugly.
JD
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56306 is a reply to message #56232] |
Thu, 11 October 2007 09:47 |
R.G.Stockwell
Messages: 163
Registered: October 2004
|
Senior Member |
|
|
"Paul van Delst" <Paul.vanDelst@noaa.gov> wrote in message
news:feliu2$p9t$1@news.nems.noaa.gov...
...
> Goodness.... I never even thought of that. Why on Earth should the order
> of the case selectors matter in a CASE construct? I just wrote:
>
> pro test_case, number
> case number of
> else: print, 'I do not like this number!'
> 1: print, '1 is a good number'
> 2: print, '2 is even better'
> 3: print, '3 is weird looking'
> endcase
> end
That seems exactly like it should behave to me.
It executes the first match "else' and jumps out.
I figure the code above is something like
pro myjunk
print,'hi'
return
print, 'never executed'
end
Cheers,
bob
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56307 is a reply to message #56306] |
Thu, 11 October 2007 09:42 |
Paul Van Delst[1]
Messages: 1157
Registered: April 2002
|
Senior Member |
|
|
Paul van Delst wrote:
> Jean H wrote:
>>> You brought up something interesting (to me anyway.) If you place
>>> the
>>> ELSE earlier in the SWITCH statement, then everything below it is
>>> executed. If you do the same for a CASE statement then only then
>>> ELSE is executed. It never occurred to me until I saw your post
>>> that it was possible to place the ELSE anywhere but at the end.
>>
>> This is useless and can bring serious headaches...
>>
>> Both the Case and Switch run from the 1st condition to the last one...
>> so anything you write below the "else" will never (Case) or always
>> (Switch) be executed... then there is no need for a Case or Switch
>> anymore! ... just delete (Case) your code or write it outside of the
>> Switch block!
>
> Goodness.... I never even thought of that. Why on Earth should the order
> of the case selectors matter in a CASE construct? I just wrote:
>
> pro test_case, number
> case number of
> else: print, 'I do not like this number!'
> 1: print, '1 is a good number'
> 2: print, '2 is even better'
> 3: print, '3 is weird looking'
> endcase
> end
>
> IDL> .run test_case
> % Compiled module: TEST_CASE.
> IDL> test_case, 1
> I do not like this number!
> IDL> test_case, 3
> I do not like this number!
> IDL> test_case, 2
> I do not like this number!
> IDL> test_case,1000
> I do not like this number!
>
> That is just ridiculous behaviour.
Ruby forces you to put it last:
#!/usr/bin/env ruby
number=ARGV[0].to_i
case number
else
puts("I don't like #{number}")
when 1..3
puts("#{number} is a good number")
end
lnx:scratch : ruby test_case.rb 5
test_case.rb:4: syntax error, unexpected kELSE, expecting kWHEN
test_case.rb:6: syntax error, unexpected kWHEN, expecting $end
when 1..3
^
In the right order it works as expected:
#!/usr/bin/env ruby
number=ARGV[0].to_i
case number
when 1..3
puts("#{number} is a good number")
else
puts("I don't like #{number}")
end
lnx:scratch : ruby test_case.rb 1
1 is a good number
lnx:scratch : ruby test_case.rb 5
I don't like 5
Ideally I would prefer the order to not matter, but if it does then I think IDL should
force the ELSE to be the last selector in CASE constructs (and probably SWITCH ones too)
cheers,
paulv
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56308 is a reply to message #56307] |
Thu, 11 October 2007 09:27 |
Paul Van Delst[1]
Messages: 1157
Registered: April 2002
|
Senior Member |
|
|
Jean H wrote:
>> You brought up something interesting (to me anyway.) If you place
>> the
>> ELSE earlier in the SWITCH statement, then everything below it is
>> executed. If you do the same for a CASE statement then only then
>> ELSE is executed. It never occurred to me until I saw your post
>> that it was possible to place the ELSE anywhere but at the end.
>
> This is useless and can bring serious headaches...
>
> Both the Case and Switch run from the 1st condition to the last one...
> so anything you write below the "else" will never (Case) or always
> (Switch) be executed... then there is no need for a Case or Switch
> anymore! ... just delete (Case) your code or write it outside of the
> Switch block!
Goodness.... I never even thought of that. Why on Earth should the order of the case
selectors matter in a CASE construct? I just wrote:
pro test_case, number
case number of
else: print, 'I do not like this number!'
1: print, '1 is a good number'
2: print, '2 is even better'
3: print, '3 is weird looking'
endcase
end
IDL> .run test_case
% Compiled module: TEST_CASE.
IDL> test_case, 1
I do not like this number!
IDL> test_case, 3
I do not like this number!
IDL> test_case, 2
I do not like this number!
IDL> test_case,1000
I do not like this number!
That is just ridiculous behaviour.
The Fortran95 equivalent:
program test_case
integer :: number
write(*,'(/,"Enter a number:")',advance='no')
read(*,'(i10)') number
select case (number)
case default
write(*,*) 'I do not like this number!'
case (1:3)
write(*,*) number,' is a good number!'
end select
end program test_case
doesn not depend on the order of the "CASE"s:
lnx:scratch : lf95 --f95 test_case.f90
Encountered 0 errors, 0 warnings in file test_case.f90.
lnx:scratch : a.out
Enter a number:3
3 is a good number!
lnx:scratch : a.out
Enter a number:1000
I do not like this number!
lnx:scratch : a.out
Enter a number:2
2 is a good number!
that is what I would expect to happen. Seems like the IDL case construct cribs code from
the switch construct. Tsk tsk.
cheers,
paulv
|
|
|
|
| Re: Bug in SWITCH - ELSE: statement? [message #56313 is a reply to message #56308] |
Thu, 11 October 2007 08:39 |
larkn10
Messages: 10
Registered: July 2006
|
Junior Member |
|
|
Hi Chris,
My understanding of SWITCH is that as soon as one option is true every
subsequent option is executed. The ELSE is the default if there is no
option that works, but it is always executed in a SWITCH statement.
You brought up something interesting (to me anyway.) If you place
the
ELSE earlier in the SWITCH statement, then everything below it is
executed. If you do the same for a CASE statement then only then
ELSE is executed. It never occurred to me until I saw your post
that it was possible to place the ELSE anywhere but at the end.
Anyway, I don't know if this is helpful.
EXAMPLE:
--------
pro switchtest
a = 5
switch a of
else:print,'switchelse'
1:print,'no1'
3:print,'no3'
5:print,'yes5'
6:print,'6'
else:print,'switchelse'
endswitch
case a of
else:print,'caseelse'
1:print,'case1'
5:print,'case5'
endcase
end;switchtest
OUTPUT:
-------
switchelse
no1
no3
yes5
6
switchelse
caseelse
-Larry
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
