forums.silverfrost.com Forum Index forums.silverfrost.com
Welcome to the Silverfrost forums
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Request undefined variables be named when detected

 
Post new topic   Reply to topic    forums.silverfrost.com Forum Index -> Suggestions
View previous topic :: View next topic  
Author Message
mecej4



Joined: 31 Oct 2006
Posts: 1430

PostPosted: Sat Feb 15, 2020 2:17 pm    Post subject: Request undefined variables be named when detected Reply with quote

In error messages following the detection of an undefined variable being used, when /checkmate was used to compile the program being run, only the line numbers and routine names are printed out. It would be very helpful if the particular variable responsible were named in the error message.

Here is an example. The buggy line of code:
Code:
IF (KA.LE.N-M+1) GA(KA+M-1)=GA(KA+M-1)+0.25D 0*PAR*(+A1*GA3(5)-A2*GA4(4))


Here is the error message that is printed at present when the bug is detected and the program is terminated:

Code:
Reference to undefined variable, array element or function result (/UNDEF) at address 5c767f

Within file teql.exe
in TAGU18 in line 5815, at address 22ad5

...more traceback lines...


Contrast this with the message from NAG Fortran for the same bug:

Code:
Runtime Error: tssubs.f, line 5815: Reference to undefined variable GA(KA+M-1)


Thanks.
Back to top
View user's profile Send private message
PaulLaidler
Site Admin


Joined: 21 Feb 2005
Posts: 6593
Location: Salford, UK

PostPosted: Sat Feb 15, 2020 4:38 pm    Post subject: Reply with quote

Thanks for the feedback. I have made a note of this.
Back to top
View user's profile Send private message AIM Address
PaulLaidler
Site Admin


Joined: 21 Feb 2005
Posts: 6593
Location: Salford, UK

PostPosted: Tue Feb 25, 2020 2:19 pm    Post subject: Reply with quote

This suggestion has been adopted and work is in progress. The aim is to implement this for 64 bits and notification for simple variables is already in place (for the next release).
Back to top
View user's profile Send private message AIM Address
John-Silver



Joined: 30 Jul 2013
Posts: 1451
Location: Aerospace Valley

PostPosted: Fri Sep 25, 2020 2:10 pm    Post subject: Reply with quote

Corrected for a myriad of spelling/typing errors on 5/10/2020 in case that was holding back on getting a reply.

Paul,back in February you wrote:
Quote:
This suggestion has been adopted and work is in progress. The aim is to implement this for 64 bits and notification for simple variables is already in place (for the next release).


I've noted over the last year or so this type of selective mod to 64bit only has been referred to more and more often.

Can I just ask whats the specific logic used to decide what to update for 32 bit, and what for 64bit, or both ?

My intiuition tells me to prepare myself for an imminent 'bucketing', or at least 'sidelining' (with for example not all new/enhanced capabilities being included in both 32 and 64 bit versions).

This approach has a certain logic, I guess, but it's a little disconcerting (for me at least) to see what appears to be a sidelining of a very stable part of the Ftn05 family.

Resoucres are of course limited, but is there , bearing in mind the still (imo) somewhat 'uncertain' nature of 64 bit ftn95, a well-founded drive to the sidelining of the 32bit version ?

It's just that it' useful for comitted 32 bit users to know in advance.

I suspect Eddie might be top of that list, and although he's the last person I'd even think about not being able to adapt to a uniquely 64bit world,he's dedicated to the 32bit for a very good reason - it's stability.
_________________
''Computers (HAL and MARVIN excepted) are incredibly rigid. They question nothing. Especially input data.Human beings are incredibly trusting of computers and don't check input data. Together cocking up even the simplest calculation ... Smile "
Back to top
View user's profile Send private message
DanRRight



Joined: 10 Mar 2008
Posts: 2235
Location: South Pole, Antarctica

PostPosted: Tue Oct 27, 2020 6:09 pm    Post subject: Reply with quote

I do not keep 32bit version on sidelines for almost a year. Before if tricky error appeared i had no choice but scale down 64bits to 32buts. But bit by bit 64bit started to beat 32bit a bit and do not dig into 32th basement floor anymore.
I understand that for a while the 32 and 64 bit version will be supported because there is no reason to switch to 64 if 32 works. I do not know how painful this synchronization job for Silverfrost is. On my experience on my own programs supporting two versions always was a nighmire hell. If for Silverftost this also is painful then my suggestion is to perform long overdue optimization of just the 64bit version to reach Intel IVF on Polyhedron tests and all users will switch to 64bits overnight. A lot of new users will also join or switch to Silverfrost i expect as this is the major issue which drags FTN95 from much broader adoption. Adding parallelization will be almost all what people need. Good also probably to do some cosmetic lift and change the name to FTNX or something
Back to top
View user's profile Send private message
PaulLaidler
Site Admin


Joined: 21 Feb 2005
Posts: 6593
Location: Salford, UK

PostPosted: Tue Oct 27, 2020 8:12 pm    Post subject: Reply with quote

Dan

Thanks for the feedback.
Back to top
View user's profile Send private message AIM Address
Display posts from previous:   
Post new topic   Reply to topic    forums.silverfrost.com Forum Index -> Suggestions All times are GMT + 1 Hour
Page 1 of 1

 
Jump to:  
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