Hi Mauro, Laurent,
I can see this thread didn't follow up.
Regarding Mauro's reasons about why not switching to meson, I totally
agree with this response from Laurent's. I don't see any of those
aspects being really an argument on why not supporting meson, but in any
case, discussion can follow up from there.
On 3/20/21 5:01 PM, Laurent Pinchart wrote:
Hi Mauro,
On Sat, Mar 20, 2021 at 08:32:09PM +0100, Mauro Carvalho Chehab wrote:
[snip]
I'm not a lover of autoconf tools. Yet, replacing from it on this
project sounds a bad idea, for a couple of reasons.
The main one is that nobody has yet provided any real reason about
*why* auto-tools should be replaced.
IMHO, switching from autotools to meson has many advantages, like a
clear syntax, an active community and *faster* build speed than autotools.
On a cold cache:
$ time (meson setup build/ -Ddefault_library=both ; ninja -j4 -C build/)
real 0m15.945s ; user 0m51.235s ; sys 0m6.914s
$ time (./configure ; make -j4)
real 0m45.985s ; user 1m57.454s ; sys 0m9.404s
Although I haven't been involved in v4l-utils previously, I can see some
people interested on having meson build support, I'm not the only one
involved in this patchset.
I understand that every change comes with a cost, but I believe meson's
been adopted for some time now, other projects have already done the
move, and the advantages overcome the possible negative aspects
mentioned. Again, this is just my humble opinion, and to me that's the
*why* porting to meson is a good idea :-)
I'd like to hear for other reasons.
Thanks,
Ariel
