 |
forums.silverfrost.com
Welcome to the Silverfrost forums
|
| View previous topic :: View next topic |
| Author |
Message |
brucebowler
Guest
|
 Posted: Tue Nov 27, 2012 4:40 pm Post subject: namelist regression? |
 |
|
I have some code that has "just worked" "forever". I haven't had to run it in a couple of years, but now find that I need to run it again. Last time I ran it, the installed version of ftn95 was 5.{something}, I'm now running 6.3.
My namelist definition looks like this. | Code: | namelist /optlist/ &
ac9Cals, ac9ScatterType, &
ac9pretime,ac9posttime, ac9PreTemp, ac9PostTemp, ac9prea,ac9prec,ac9posta,ac9postc,&
ac9NumGooses,ac9Gooses, &
dawnCals, dawnpretime,dawnposttime,dawnprecal,dawnpostcal, &
chlCals, chlType, chlStart, chlEnd, chlCoeffCount, chlCoeff, &
hsNumSweeps, hsSweeps, &
gmtoffsethours, hscatinifile, sasinifile, firstStation
|
My input file looks like this...
| Code: | &optlist
chlCoeff(1,1:3) = -100602.0067, 620.2207, -0.9559
chlType = 3
chlStart = 323.8
chlEnd = 324.6
chlCoeffCount = 3
chlCals = 1
/
|
My reading code looks like this...
| Code: | open (unit=optunit,file=filename,status='readonly')
read (optunit, nml=optlist)
close (unit=optunit)
|
When I run it, I get an "Error 204, invalid identifier in namelist input group"
The problem appears to be related to chlCoeff and chlCoeffCount. If I remove the 'e' from those names in both the namelist definition, declarations, and input file, everything works fine. I haven't tried changing the 'e' to some other letter to see if it's specifically the 'e', but I'd really like to leave the 'e' in there so I don't have to go and change roughly 1000 input files when I need to reprocess old data.
Any idea as to why the failure?
PS, there's an implicit none at the start of the code and all the variables have been properly defined. In particular, | Code: | real chlCoeff(10,10)
integer chlCoeffCount |
|
|
| Back to top |
|
 |
mecej4
Joined: 31 Oct 2006
Posts: 1899
|
| Posted: Tue Nov 27, 2012 5:32 pm Post subject: Re: namelist regression? |
|
|
The reported error could not be reproduced using the code fragments posted and using implicit typing for variables whose declarations were not shown.
The complete test source needs to be shown to enable one to reproduce the reported behavior. |
|
| Back to top |
|
|
PaulLaidler
Site Admin
Joined: 21 Feb 2005
Posts: 8210
Location: Salford, UK
|
| Posted: Tue Nov 27, 2012 5:53 pm Post subject: |
|
|
It works OK for me so the first thing to check is that you are not accessing an old salflibc.dll by mistake.
I am running 64 bit Windows 7. |
|
| Back to top |
|
|
brucebowler
Guest
|
| Posted: Tue Nov 27, 2012 6:54 pm Post subject: |
|
|
Only salflibc.dll's on the machine are in c:\program files\ftn95 and \redist. Both are dated 5/14/2012 and are 2007KB. Internal version for both is 14.5.14.22.
Original code is 1000's of lines long, but I will attempt to make a small reproducer in the mean time.
Bruce |
|
| Back to top |
|
|
brucebowler
Guest
|
| Posted: Tue Nov 27, 2012 7:51 pm Post subject: |
|
|
As presented, this works. If you do a global search and replace of Coff with Coeff, it fails for me.
| Code: | program testing
implicit none
character (len=128) :: optfilename
integer :: optlen
optfilename = 'v:\gnats\2012\s121118e\s121118e.opt'
optlen = leng(optfilename)
call readOptFile(optfilename(1:optlen))
end
subroutine readOptFile(filename)
implicit none
character (len=*), intent(in) :: filename
integer :: ac9ScatterType, ac9Cals,ac9NumGooses
real :: ac9PreTime,ac9PostTime, ac9PreTemp,ac9PostTemp
real :: ac9PreA(9),ac9PreC(9),ac9PostA(9),ac9PostC(9)
real :: ac9Gooses(10,3)
integer:: dawnCals
real :: dawnPreTime,dawnPostTime, dawnPreCal, dawnPostCal
integer :: chlType(10), chlCoffCount(10), chlCals
real :: chlStart(10), chlEnd(10), chlCoff(10,10)
integer :: hsNumSweeps
real*8 :: hsSweeps(10,3)
integer :: gmtoffsethours,firstStation
character (len=128) :: hscatinifile, sasinifile
integer optunit
namelist /optlist/ &
ac9Cals, ac9ScatterType, ac9pretime,ac9posttime, ac9PreTemp, ac9PostTemp, &
ac9prea,ac9prec,ac9posta,ac9postc,ac9NumGooses,ac9Gooses, &
dawnCals, dawnpretime,dawnposttime,dawnprecal,dawnpostcal, &
chlCals, chlType, chlStart, chlEnd, chlCoffCount, chlCoff, &
hsNumSweeps, hsSweeps, gmtoffsethours, hscatinifile, sasinifile, firstStation
optunit=15
open (unit=optunit,file=filename,status='readonly')
read (optunit, nml=optlist)
close (unit=optunit)
return
end |
.opt file
| Code: | &optlist
chlCoff(1,1:3) = -100602.0067, 620.2207, -0.9559
chlType = 3
chlStart = 323.8
chlEnd = 324.6
chlCoffCount = 3
chlCals = 1
/ |
|
|
| Back to top |
|
|
davidb
Joined: 17 Jul 2009
Posts: 560
Location: UK
|
|
| Back to top |
|
|
brucebowler
Guest
|
| Posted: Tue Nov 27, 2012 8:16 pm Post subject: |
|
|
Yep, saw that (and even commented on it back then), but I don't think it's related...
edit=== rereading, yes, I think it is related.
Paul, any chance of getting this looked at?
Last edited by brucebowler on Tue Nov 27, 2012 8:50 pm; edited 1 time in total |
|
| Back to top |
|
|
davidb
Joined: 17 Jul 2009
Posts: 560
Location: UK
|
| Posted: Tue Nov 27, 2012 8:34 pm Post subject: |
|
|
If I run "my" code from the other post with the name PROGNAME for one of the items in the namelist, I get the same error as you.
Error 204, invalid identifier in namelist input group
Edit. Re-reading the posts on the other thread it looks like this Error 204 was reported back then. There is also this post from someone called Martin (presumably from Silverfrost), who reproduced the behaviour.
http://forums.silverfrost.com/viewtopic.php?p=7681&sid=45a3a42c1f90c62bd843d75c11e74430
It seems that some names are just not liked, though I haven't figured out what the pattern is.
I get a similar error with another compiler (won't mention which one), which reports erroneous ',' when reading in the namelist. Its all a bit strange.
_________________
Programmer in: Fortran 77/95/2003/2008, C, C++ (& OpenMP), java, Python, Perl
Last edited by davidb on Tue Nov 27, 2012 9:37 pm; edited 2 times in total |
|
| Back to top |
|
|
brucebowler
Guest
|
| Posted: Tue Nov 27, 2012 8:45 pm Post subject: |
|
|
| I know my code fails in 6.0 and 6.3 (and I presume those in between) and worked in 5.x (I just don't recall which values of x :-) |
|
| Back to top |
|
|
PaulLaidler
Site Admin
Joined: 21 Feb 2005
Posts: 8210
Location: Salford, UK
|
| Posted: Wed Nov 28, 2012 9:53 am Post subject: |
|
|
There was a bug that Martin reported as fixed and this appears to be the case on some operating systems. If you can provide a complete but short sample program together with details of your operating system then we can take another look at it.
David, could you verify that you are not accessing an old salflibc.dll. |
|
| Back to top |
|
|
brucebowler
Guest
|
| Posted: Wed Nov 28, 2012 12:55 pm Post subject: |
|
|
See complete short example above. If that's not short enough, I'll try and make it shorter.
Win XP, service pack 3, 32bit. up to date with all windows patches etc as of yesterday. multi-processor "hardware" (it's really a VMware virtual machine) |
|
| Back to top |
|
|
PaulLaidler
Site Admin
Joined: 21 Feb 2005
Posts: 8210
Location: Salford, UK
|
| Posted: Wed Nov 28, 2012 4:00 pm Post subject: |
|
|
Your sample runs OK for me under Windows 7 and XP.
I have changed the names to include the e.
I thought that the fact that chlCoeffCount is declared as an array may be a problem but the program executes to the end either way.
I will check to see if the namelist bug fix was made after the last release. |
|
| Back to top |
|
|
davidb
Joined: 17 Jul 2009
Posts: 560
Location: UK
|
| Posted: Wed Nov 28, 2012 7:14 pm Post subject: |
|
|
Hi Paul,
I am accessing the DLL that is in the install folder. The file version is 14.5.14.22. Date modified is 14/05/2012.
This is the short example I have the same problem with. I'm running Windows Vista 32bit and FTN95 6.3.
| Code: |
PROGRAM MAIN
INTEGER MAXLINES, I
PARAMETER(MAXLINES=5)
CHARACTER*100 ASSOCIATE,PROGNAME,TITLE,ROWHEAD,LINE(MAXLINES)
NAMELIST /MENUSLIST/ ASSOCIATE,PROGNAME,TITLE,ROWHEAD,LINE
! Set all character arrays to values
ASSOCIATE = 'A'
PROGNAME = 'B'
TITLE = 'C'
ROWHEAD = 'D'
LINE = 'E'
OPEN(99,FILE='MENUSFILE')
WRITE(99,MENUSLIST)
CLOSE(99)
! corrupt variables
ASSOCIATE = 'X'
PROGNAME = 'X'
TITLE = 'X'
ROWHEAD = 'X'
LINE = 'X'
! Read file back in and print
OPEN(99,FILE='MENUSFILE')
READ(99,MENUSLIST)
CLOSE(99)
PRINT *,ASSOCIATE
PRINT *,PROGNAME
PRINT *,TITLE
PRINT *,ROWHEAD
DO I=1,MAXLINES
PRINT *,LINE(I)
END DO
END PROGRAM MAIN
|
_________________
Programmer in: Fortran 77/95/2003/2008, C, C++ (& OpenMP), java, Python, Perl
Last edited by davidb on Thu Nov 29, 2012 8:27 am; edited 1 time in total |
|
| Back to top |
|
|
mecej4
Joined: 31 Oct 2006
Posts: 1899
|
| Posted: Thu Nov 29, 2012 1:28 am Post subject: Re: |
|
|
| DavidB's example (self-contained, write/read namelist) exhibits the problem on Windows 7 X64 SP1, as well, when FTN95 6.30 is used. |
|
| Back to top |
|
|
DanRRight
Joined: 10 Mar 2008
Posts: 2923
Location: South Pole, Antarctica
|
| Posted: Thu Nov 29, 2012 7:03 pm Post subject: |
|
|
| And while fixing this, can additional option be made not to convert printed variables to capital letters? NowItKillsTheIdea_of_SelfExplanatoryCode |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
