forums.silverfrost.com

[DanRRight]

I get a lot of warnings of truncated lines in the lines where exclamation mark is posted even after 72 position (!) like in example here

[davidb]

To be fair, you shouldn't put anything after column 72, even a comment.

(in the old days you could put a sequence number in there in case you dropped your cards).

You can remove the 72 column limit using the option /WIDE_SOURCE

Or you can leave the 72 column limit in place and just disable the warning by using the /NO_WARN_WIDE option.

You will have to play and see which works for you. Personally, I get my editor to check for "long lines" so I never have this problem at compile time.
_________________
Programmer in: Fortran 77/95/2003/2008, C, C++ (& OpenMP), java, Python, Perl

[DanRRight]

All these workarounds are an invitation for disaster, David. Formally, we are allowed to put comment with exclamation mark in any place, so the compiler should treat that correctly.

This is large legacy code, it is not mine, it uses 72 char default limit and i permanently catch errors associated with over-the-limit use of variables with this specific compiler as the code evolves since the authors own ones (many other compilers) do not care about this and a lot of other potential bugs (as we well know that from polyhedron site). As a result instead of checking 1-2 real warning instances i have to look at hundreds. This is because just one such error -- and you will never find why the results are bad. So these warning could be deadliest errors.

Besides the compiler reports truncated lines with extreme ambiguity. Look at the code (use tab before print*)

[PaulLaidler]

I have logged this for investigation.

[DanRRight]

That over-reporting of real breakouts of 72 character limit literally kills me. There are thousands of them FTN95 reports in the code i try to use, most are wrong ones caused by incorrect treatment of line width with exclamation mark used. The F77 compiler (or G77 i do not remember) code authors use is so bad that it does not stops even if it you get over the limit with such statements cutting some letters in THEN like in the ecxample line here

IF (asddsfgdfgdfgdffdsdfsdf.GT.sdfdfgdfdfgdfggdfgdfg (sdfdfgdfgddfdfdfdf+1) THEN

It does not stop if the variable is in implicit none list and is cut. It does not stop with the line missing the end of character variable....total disaster. I have tried to push them to use FTN95 but you all know it is impossible to move people to use other compilers. So, Paul, please look at this ASAP, years go and i still can not make the code working with this compiler. I can not use other compiler for this code when i see how dumb and buggy this POS is. Clearly, such bad compilers are not usable for large codes

[LitusSaxonicum]

It looks to me like a conflict between the checks done for Fortran 77 and earlier, with the checks done for Fortran 90 and later. In Fortran 77 we don't have the ! comment, so going over 72 columns is forbidden, whereas in Fortran 90 with a ! comment the statement goes on until the CR LF sequence.

Perhaps the logical thing is to consider a statement as terminated by a ! comment for length computation purposes, as the comment cannot be followed by executable statements on the same line because ; would be considered part of the comment - rather than considering the statement to be extended to the end of the comment, which is what appears to be done.

I've been putting & in column 73 and changing my continuation characters to & so that my code is future-proof, and & in column 73 doesn't seem to trigger the adverse reaction that Dan has received.

Eddie

[DanRRight]

This problem solution is simple. With 72 or 132 char limits the line should not be considered truncated if any letters in the line end within these limits AND the ! sign. Spaces and tabs between last letter and ! sign should not be considered as violators of limits opposed to what seems is going on right now. Well, unless i do not see some pitfalls from Fortran66 ages

Thanks for hints to make future-proof sources, Eddie, i was also experimenting somewhere in this direction.

[LitusSaxonicum]

Not an original idea from me, just following what I saw in the INC files. As usual, someone at Silverfrost (or Salford) was pretty smart. I do miss my 1,2,3 ... and a,b,c ... continuations, but I don't think I'll every drop a box full of 80 column cards again (like I once did on the 5th floor of a staircase with open treads!).

Eddie

[PS - I don't write code like this, but when it is working, I tidy it up so that the programs compile in fixed or free format. I don't go over the 80 columns of a card, as on A4 paper I can print smaller, but then I can't read it! The column 7 start and statement numbers look like 'outdents' to me - which I like]

[DanRRight]

I see GNU Fortran neglects exclamation marks as comments in Format statements