8:01 a.m.
Based on discussion with Eric:
https://www.redhat.com/archives/libvir-list/2014-March/msg01001.html
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
v2:
- rebase on current master and s/FILENAME:1:/FILENAME ":1:"/
cfg.mk | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/cfg.mk b/cfg.mk
index 675af21..80440b5 100644
--- a/cfg.mk
+++ b/cfg.mk
@@ -938,6 +938,13 @@ sc_require_locale_h:
fi; \
done;
+sc_prohibit_empty_first_line:
+ @awk 'BEGIN { fail=0; } \
+ FNR == 1 { if ($$0 == "") { print FILENAME ":1:"; fail=1; } } \
+ END { if (fail == 1) { \
+ print "$(ME): Prohibited empty first line" > "/dev/stderr"; \
+ } exit fail; }' $$($(VC_LIST_EXCEPT) | grep '\.[ch]$$');
+
# We don't use this feature of maint.mk.
prev_version_file = /dev/null
--
2.0.0
10:55 a.m.
On 04.06.2014 15:01, Martin Kletzander wrote:
Based on discussion with Eric:
https://www.redhat.com/archives/libvir-list/2014-March/msg01001.html
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
v2:
- rebase on current master and s/FILENAME:1:/FILENAME ":1:"/
cfg.mk | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/cfg.mk b/cfg.mk
index 675af21..80440b5 100644
--- a/cfg.mk
+++ b/cfg.mk
@@ -938,6 +938,13 @@ sc_require_locale_h:
fi; \
done;
+sc_prohibit_empty_first_line:
+ @awk 'BEGIN { fail=0; } \
+ FNR == 1 { if ($$0 == "") { print FILENAME ":1:"; fail=1; } }
\
+ END { if (fail == 1) { \
+ print "$(ME): Prohibited empty first line" > "/dev/stderr";
\
+ } exit fail; }' $$($(VC_LIST_EXCEPT) | grep '\.[ch]$$');
Maybe we can check the whole VC_LIST, there are only a few files that
start with an empty line (typically README.txt) and those can have an
exception. But there are some that shouldn't start with an empty line, e.g.:
docs/generic.css:1:
src/locking/lockd.conf:1:
src/locking/sanlock.conf:1:
and some testdata under
tests/nodeinfodata/linux-test3/node/node*/meminfo. These in particular
should be plain copy of their origins in sysfs which at least on my
system doesn't start with an empty line:
# head -n1 /sys/devices/system/node/node0/meminfo
Node 0 MemTotal: 4004132 kB
But your version is good too. ACK whether you decide to change it the
way I'm suggesting or not.
Michal
11:14 a.m.
On 06/04/2014 07:01 AM, Martin Kletzander wrote:
Based on discussion with Eric:
https://www.redhat.com/archives/libvir-list/2014-March/msg01001.html
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
v2:
- rebase on current master and s/FILENAME:1:/FILENAME ":1:"/
cfg.mk | 7 +++++++
1 file changed, 7 insertions(+)
The existing sc_prohibit_empty_lines_at_EOF in maint.mk does a similar
check but using perl instead of awk. But that shouldn't matter.
diff --git a/cfg.mk b/cfg.mk
index 675af21..80440b5 100644
--- a/cfg.mk
+++ b/cfg.mk
@@ -938,6 +938,13 @@ sc_require_locale_h:
fi; \
done;
+sc_prohibit_empty_first_line:
+ @awk 'BEGIN { fail=0; } \
+ FNR == 1 { if ($$0 == "") { print FILENAME ":1:"; fail=1; } } \
why is this to stdout...
+ END { if (fail == 1) { \
+ print "$(ME): Prohibited empty first line" > "/dev/stderr"; \
...but this to stderr?
+ } exit fail; }' $$($(VC_LIST_EXCEPT) | grep
'\.[ch]$$');
I'd use ALL of VC_LIST_EXCEPT (not just the .c and .h files). But in
general it looks like it does the right thing.
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
4:59 a.m.
On Wed, Jun 04, 2014 at 10:14:24AM -0600, Eric Blake wrote:
On 06/04/2014 07:01 AM, Martin Kletzander wrote:
> Based on discussion with Eric:
>
> https://www.redhat.com/archives/libvir-list/2014-March/msg01001.html
>
> Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
> ---
> v2:
> - rebase on current master and s/FILENAME:1:/FILENAME ":1:"/
>
> cfg.mk | 7 +++++++
> 1 file changed, 7 insertions(+)
The existing sc_prohibit_empty_lines_at_EOF in maint.mk does a similar
check but using perl instead of awk. But that shouldn't matter.
Yeah, but that is taken from gnulib, isn't it? Do you suggest
proposing this fix there?
>
> diff --git a/cfg.mk b/cfg.mk
> index 675af21..80440b5 100644
> --- a/cfg.mk
> +++ b/cfg.mk
> @@ -938,6 +938,13 @@ sc_require_locale_h:
> fi; \
> done;
>
> +sc_prohibit_empty_first_line:
> + @awk 'BEGIN { fail=0; } \
> + FNR == 1 { if ($$0 == "") { print FILENAME ":1:"; fail=1; } }
\
why is this to stdout...
I wanted to simulate the behaviour of other syntax-check rules that
use grep without output redirection. No problem of making this go to
stderr, though.
> + END { if (fail == 1) {
\
> + print "$(ME): Prohibited empty first line" >
"/dev/stderr"; \
...but this to stderr?
> + } exit fail; }' $$($(VC_LIST_EXCEPT) | grep '\.[ch]$$');
I'd use ALL of VC_LIST_EXCEPT (not just the .c and .h files). But in
general it looks like it does the right thing.
Michal had a good point of those other files. I'll remove the grep,
add necessary files into except rule and fix all the other non-.[ch]
ones.
Thanks for the reviews
Martin
3823
days inactive
3824
days old
3 comments
3 participants
participants (3)
-
Eric Blake -
Martin Kletzander -
Michal Privoznik