[Fortran] Massive whitespace patch.

Brooks Moses brooks.moses@codesourcery.com
Tue Jan 2 00:30:00 GMT 2007


Steve Kargl wrote:
> On Mon, Jan 01, 2007 at 03:31:14PM -0800, Brooks Moses wrote:
>> I think what Thomas is referring to is that any patchfile for 4.3 that 
>> touches a spot with changed whitespace will not apply to 4.2, and will 
>> thus require hand editing of the files to backport.  This makes 
>> backporting things substantially more laborious than the "apply the 
>> patch and regtest" process that it is now.
> 
> I'm well aware of this complication.  But, I stand by my assertation
> that the whitespace patch does not prevent back porting.

And I am not disputing this assertion -- "substantially more laborious" 
is a very different thing from "impossible".  :)

I would be somewhat curious to know approximately how much of the 
codebase this ends up touching, on a percentage-of-lines-of-code basis; 
is this something like one line out of twenty, or more like every other 
line?

>> (My personal opinion is that this annoyance and extra work does not 
>> outweigh the value of doing a style cleanup.  I'm not sure about timing, 
>> though -- I can see arguments for why it might be useful to hold off 
>> until the 4.2 release, but I don't have any opinion one way or the other 
>> on their validity.)
> 
> The timing is such that I will not work on the ISO C Binding 
> branch until after the whitespace patch is merged into branch.

In which case, I'd agree that that's by far the most significant reason 
that I see for committing this at some particular point or another.

On a different note, how much of this work did you do manually, and how 
much by automated search-and-replace stuff?  If there was a lot of the 
latter, I think it would be helpful if you could share some of the 
techniques, because I suspect that merging this into various other 
overall-GCC development branches is going to be "interesting" if they've 
touched a fair bit of Fortran code.  (I don't know what the situation is 
what that, overall; I can say that what I'm helping with on the new 
CALL_EXPR representation from the LTO branch has some 35kb of Fortran 
patches.)

- Brooks



More information about the Fortran mailing list