| 1 |
El dom, 20-12-2015 a las 13:13 +0100, Ulrich Mueller escribió: |
| 2 |
> > > > > > On Sun, 20 Dec 2015, Michał Górny wrote: |
| 3 |
> |
| 4 |
> > > Besides, the "set -f" code in the eclass works just fine and has |
| 5 |
> > > been tested for several years by now. |
| 6 |
> |
| 7 |
> > Sure. Relying on implicit word splitting which is a side effect of |
| 8 |
> > unquoted variable use, along with disabling side effects of that |
| 9 |
> > use |
| 10 |
> > is not a hack. On the other hand, using built-in dedicated to word |
| 11 |
> > splitting is. Of course. |
| 12 |
> |
| 13 |
> > While at it, please make sure to remove all uses of 'find -print0' |
| 14 |
> > hack. I'm pretty sure you can find some implicit ugly side effect |
| 15 |
> > that could allow us to do the same after mangling a few shopts. |
| 16 |
> |
| 17 |
> So, you don't get things exactly in your style, and then other dev's |
| 18 |
> code is an "ugly hack" and uses "side effects". Even if the code in |
| 19 |
> question causes no problems whatsoever and uses standard well |
| 20 |
> documented shell features. |
| 21 |
> |
| 22 |
> I think we should end this discussion here. Let the eclass maintainer |
| 23 |
> decide what approach (if any) he wants to adopt. |
| 24 |
> |
| 25 |
> Ulrich |
| 26 |
|
| 27 |
I prefer to keep the original behavior as ulm patch did :| |
Archives