This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch, Fortran, OOP] PR 56284: ICE with alternate return in type-bound procedure

On 4/4/2013 1:14 PM, Janus Weil wrote:
As my editor shows, that file uses DOS line endings (\r\n) in some lines and UNIX ones (\n) in others. In principle, I am for keeping such files to test the parser.

If one keeps them, please put them into a file that tests for that feature exclusively, as we often find turds, oops, I mean bits in the tree were someone did the wrong thing and accidentally put in CRLRs.

I would also prefer to remove them, in particular since they are not
even consistently used in the whole file, but only in some lines.
Moreover it seems that there are still enough of those guys around in
the testsuite to 'test the parser' (if this is necessary).

So I will just commit the patch as posted (with Tobias' ok).

It's amazing how much effort goes into dealing with the differing
line endings. It's bad luck that Unix standardized on a premature
version of ASCII (containing ASCII.NL) instead of the final form
with ASCII.CR and ASCII.LF. Of course with modern equipment retaining
the separation of functions is a bit silly (not many people using
teletypes these days :-)) but as with all standards, it is really
annoying to have two!

I must say though that Apple choosing to use CR only was simply
perverse :-)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]