Re: [PATCH] srcpos: correct column numbers

[David Gibson <david-xT8FGy+AXnRB3Ne2BGzF6laj5H9X9Tb+@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 3689 bytes --]

On Thu, Jan 18, 2018 at 11:12:16AM +0100, Julia Lawall wrote:
> 
> 
> On Thu, 18 Jan 2018, David Gibson wrote:
> 
> > On Thu, Jan 18, 2018 at 06:15:47AM +0100, Julia Lawall wrote:
> > >
> > >
> > > On Thu, 18 Jan 2018, David Gibson wrote:
> > >
> > > > On Tue, Jan 16, 2018 at 07:10:23AM +0100, Julia Lawall wrote:
> > > > >
> > > > >
> > > > > On Mon, 15 Jan 2018, Frank Rowand wrote:
> > > > >
> > > > > > On 01/15/18 10:33, Julia Lawall wrote:
> > > > > > > The start of a line is column 0, at least according to emacs.
> > > > > >
> > > > > > According to vim, the first character of a line is column 1.  I don't know
> > > > > > if it has a concept of column 0, to the left of that character.
> > > > > >
> > > > > > $ vim --version
> > > > > > VIM - Vi IMproved 7.4
> > > > > >
> > > > > > Let the editor wars begin.... :-)
> > > > > >
> > > > > > Personally, I use vim, but if the dtc column numbers match emac's world view
> > > > > > instead of vim's, that is fine with me.
> > > > >
> > > > > At the start of a line, ie just after a \n, there has been no character.
> > > > > This change made the column numbers work for me in the annotations.
> > > > >
> > > > > There is also the issue of tabs.  David wanted tabs to count as 8
> > > > > characters, but actually cpp converts tabs to spaces.  So if a dts file
> > > > > has been through cpp, the numbers will be smaller.  So it could be better
> > > > > to consider tabs to be 1 character, to not have to think about what kind
> > > > > of include was used.
> > > >
> > > > Urgh.. debating which editor to match sounds like it's going to be a
> > > > rathole.
> > > >
> > > > Let's try a different approach: does gcc ever report column numbers?
> > > > If so, what's its interpretation?
> > >
> > > Chosen at random, I have the following error:
> > >
> > > drivers/thermal/kirkwood_thermal.c:88:15: warning: assignment makes
> > > pointer from
> > >  integer without a cast [enabled by default]
> > >   priv->sensor = devm_request_and_ioremap(&pdev->dev, res);
> > >
> > > and the following code:
> > >
> > > 	priv->sensor = devm_request_and_ioremap(&pdev->dev, res);
> > >
> > > In emacs, the go to the line and type control f 15 times solution puts the
> > > cursor right afher after the =.  The column numbers of emacs report 22 for
> > > that position.  I guess in vim, putting the cursor right on the = would
> > > report 15.
> > >
> > > When I run the dts compiler on tests/test_tree1.dts, for line 39:
> > >
> > >
> > >                         compatible = "subsubnode2", "subsubnode";
> > >
> > > I get the start and end as 24 and 65.  24 is what emacs shows as the
> > > column number when I put the cursor right on the c.  65 is what the column
> > > number shows when I put the cursor right past the semicolon.
> > >
> > > I tried to move around in vim, but I don't know how it works.  It seems to
> > > report the position of the c as 4 or 25.  It seems to report the position
> > > of the final semicolon as 44 or 65.  I am not sure how it work though,
> > > because it gives the impression that there is a single blank character
> > > before c, which is not the case.
> >
> > Ok, let's go with the gcc definition, which appears to be "# of bytes
> > past the last \n".
> 
> So that would be the strategy of the patch, where seeing a newline sets
> the counter to 0.

Right, but we should kill the tab expansion stuff at the same time.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]