On 30 April 2016 at 03:01, Alexander von Gluck IV <kallisti5@xxxxxxxxxxx> wrote:
April 29 2016 3:02 AM, "Adrien Destugues" <pulkomandy@xxxxxxxxxxxxx> wrote:
29 avril 2016 09:36 "Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx> a écrit:
Am 28/04/2016 um 18:39 schrieb Adrien Destugues:
https://sourceforge.net/p/flex/bugs/140
So yes, looks like a bug in Flex. I would rather require people to
avoid the broken version, then, instead of hacking around it (one
more check to add to our configure script, maybe).
Fixed in 2012, and now it hits us. We really do live in a fast paced world
;-)
For reference, this is distros shipping the broken 2.5.37 version:
http://distrowatch.com/search.php?pkg=flex&pkgver=2.5.37&distrorange=InLatest#pkgsearch
And this is distro having already upgraded to a working version (at this
time no one packages
2.5.38 which was very short-lived with 2.6.39 released just 1.5 month after
it):
http://distrowatch.com/search.php?pkg=flex&pkgver=2.5.39&distrorange=InLatest#pkgsearch
http://distrowatch.com/search.php?pkg=flex&pkgver=2.6&distrorange=InLatest#pkgsearch
Interestingly, the Ubuntu 16.04 system I have with the breakage ships with
flex 2.6.0.
Maybe just wrap my define changes with a != gcc2 check until this tainted
flex code
rolls out of the major desktops? I'd fix the real issue if I could, but the
c is
flex-generated.