Front page | perl.perl5.porters |
Postings from September 2016
On Mon Sep 05 07:59:34 2016, davem wrote: > On Mon, Sep 05, 2016 at 02:48:40PM +0100, Dagfinn Ilmari Mannsåker > wrote: > > Dave Mitchell <davem@iabyn.com> writes: > > > > > On Sun, Sep 04, 2016 at 09:30:44PM -0700, Father Chrysostomos via > > > RT wrote: > > > Around the time this commit was pushed, win32 smokers and Jenkins > > > have > > > been failing to compile, with > > > > > > ..\toke.c(7586) : error C2105: '--' needs l-value > > > > > > However, your commit doesn't touch the affected area of code, which > > > hasn't > > > changed since 2011: > > > > > > 7584: case KEY_evalbytes: > > > 7585: PL_expect = XTERM; > > > 7586: UNIBRACK(-OP_ENTEREVAL); > > > > > > The error would seem to imply that OP_ENTEREVAL has been defined to > > > a > > > negative value, but it's been 345 in opnames.h for a month now. > > > > > > So I can't see what the problem is. > > > > UNIBRACK(f) is defined as UNIUNI3(f,0,0), which as of 9bde5622 does > > > > PL_last_lop_op = f < 0 ? -f : f; > > > > So this expands to '-345 < 0 ? --345 : -345'. Changing it to '-(f)' > > (and '(f)', for consistency) shold fix it. > > I've just done that as v5.25.4-84-g0af40c7 > > On Mon, Sep 05, 2016 at 02:55:25PM +0100, Zefram wrote: > > Dave Mitchell wrote: > > > ..\toke.c(7586) : error C2105: '--' needs l-value > > ... > > > 7586: UNIBRACK(-OP_ENTEREVAL); > > > > > > The error would seem to imply that OP_ENTEREVAL has been defined to > > > a > > > negative value, > > > > That's not how the C preprocessor works. Macro expansion can't paste > > two adjacent "-" tokens into a "--". In any case, the OP_ values > > have > > never been negative. > > I guess that's a bug in the win C compiler? Which would explain why > the > problem hasn't surfaced on other platforms. I see you have beaten me to fixing it in 0af40c757f0. -- Father Chrysostomos --- via perlbug: queue: perl5 status: pending release https://rt.perl.org/Ticket/Display.html?id=129196Thread Previous | Thread Next