Here's the diff. It's a bit longer than expected, because I took the
opportunity to expand a bit on the error-reporting engine.
The actual "new error" is the bit in var.c.
The rest:
- makes the engine record the current "parse location" while expanding
commands in a gnode.
- tests fatal_errors be
On 08/18/12 13:48, Kenneth R Westerback wrote:
On Sat, Aug 18, 2012 at 12:29:51PM +0200, Marc Espie wrote:
Our make currently "misbehaves" and doesn't treat this as a problem.
A=a
a:
echo ${A
there are several possibilities.
1/ do like gmake and treat this as an actual error.
2/ keep
On Sat, Aug 18, 2012 at 12:29:51PM +0200, Marc Espie wrote:
> Our make currently "misbehaves" and doesn't treat this as a problem.
>
> A=a
>
> a:
> echo ${A
>
>
> there are several possibilities.
> 1/ do like gmake and treat this as an actual error.
> 2/ keep the status quo
> 3/ do someth
18.08.2012 14:33 полÑзоваÑÐµÐ»Ñ "Marc Espie"
напиÑал:
>
> Our make currently "misbehaves" and doesn't treat this as a problem.
>
> A=a
>
> a:
> echo ${A
>
>
> there are several possibilities.
> 1/ do like gmake and treat this as an actual error.
> 2/ keep the status quo
>
Our make currently "misbehaves" and doesn't treat this as a problem.
A=a
a:
echo ${A
there are several possibilities.
1/ do like gmake and treat this as an actual error.
2/ keep the status quo
3/ do something else: if it's not a complete variable reference, then it
should be seen as a s
