Playing with this, going the other way and moving to just using %(describe) is good, since it vastly simplifies things.

The problem is that %(describe) is rather new, and as solving this problem means solving it on thekelleys.org.uk, which is running Debian stable, the yak has now morphed into "install latest git as a backport on Debian stable." More later.....


Simon.


On 13/09/2022 09:10, Johnny S. Lee via Dnsmasq-discuss wrote:
How about something like the following?

diff --git a/VERSION b/VERSION
index 998eb1f..29a22f8 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-$Format:%d$
+$Format:%d;%(describe)$
diff --git a/bld/get-version b/bld/get-version
index 1f51768..4ba3646 100755
--- a/bld/get-version
+++ b/bld/get-version
@@ -24,13 +24,19 @@ if which git >/dev/null 2>&1 && \
      ([ -d .git ] || grep '^gitdir:' .git >/dev/null 2>&1) && \
      git describe >/dev/null 2>&1; then
      git describe | sed 's/^v//'
-elif grep '\$Format:%d\$' $1/VERSION >/dev/null 2>&1; then
+elif grep '\$Format:%d;%(describe)\$' $1/VERSION >/dev/null 2>&1; then
      # unsubstituted VERSION, but no git available.
      echo UNKNOWN
  else
-     vers=`cat $1/VERSION | sed 's/[(), ]/,/ g' | tr ',' '\n' | grep ^v[0-9]`
+     vers=`cut -d';' -f1 $1/VERSION | sed 's/[(), ]/,/ g' | tr ','
'\n' | grep ^v[0-9]`
+     ret=$?

-     if [ $? -eq 0 ]; then
+     if [ ${ret} -ne 0 ]; then
+         vers=`cut -d';' -f2 $1/VERSION | grep ^v[0-9]`
+         ret=$?
+     fi
+
+     if [ ${ret} -eq 0 ]; then
           echo "${vers}" | sort -k1.2,1.5Vr -k1.6,1.6 -k1.8,1.9Vr
-k1.10,1.11Vr | head -n 1 | sed 's/^v//'
       else
           cat $1/VERSION

This appends ";%(describe)" to the format.
The script will try to parse the contents of %d first, then %(describe).
This should have minimal impact.

On Tue, 13 Sept 2022 at 06:52, Johnny S. Lee <_...@jsl.io> wrote:

I'm not clear what the difference between that and %(describe) is. The
substituted value is used subtly in get-version as part of the
build-process, so we need to be a little careful about changing things.

$ git log -10 --format="'%d'"
' (HEAD -> master, origin/master, origin/HEAD)'
' (tag: v2.87rc1)'
''
''
''
''
''
''
''
' (tag: v2.87test9)'
$ git log -10 --format="'%(describe)'"
'v2.87rc1-1-gc4b9bc6'
'v2.87rc1'
'v2.87test9-7-g04cc2ae'
'v2.87test9-6-g32588c7'
'v2.87test9-5-g84a6d07'
'v2.87test9-4-gd6c69f6'
'v2.87test9-3-gce37291'
'v2.87test9-2-g09d741f'
'v2.87test9-1-g0666ae3'
'v2.87test9'

Output of get-version should be the same for tagged commits.
Note that the %(describe) placeholder is rather new. It was added in
Git 2.32 last year.


On Mon, 12 Sept 2022 at 23:19, Simon Kelley <si...@thekelleys.org.uk> wrote:


The current format is %d, which is

"ref names, like the --decorate option of git-log[1]"

according to git help log.

I'm not clear what the difference between that and %(describe) is. The
substituted value is used subtly in get-version as part of the
build-process, so we need to be a little careful about changing things.

When the build runs on a fully-fledged git repo, rather than a tar file
extracted from one, get-version does run "git describe". This isn't the
case during the automated build, however.


Simon.


On 12/09/2022 04:51, Johnny S. Lee via Dnsmasq-discuss wrote:
The contents of file VERSION from the tar.gz files of the latest 3 commits:

dnsmasq-c4b9bc6/VERSION: (HEAD -> master)
dnsmasq-4447d48/VERSION: (tag: v2.87rc1, origin/master, origin/HEAD)
dnsmasq-04cc2ae/VERSION:

Would it be better adding "%(describe)" to the format in file VERSION?

_______________________________________________
Dnsmasq-discuss mailing list
Dnsmasq-discuss@lists.thekelleys.org.uk
https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss


_______________________________________________
Dnsmasq-discuss mailing list
Dnsmasq-discuss@lists.thekelleys.org.uk
https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss

_______________________________________________
Dnsmasq-discuss mailing list
Dnsmasq-discuss@lists.thekelleys.org.uk
https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss


_______________________________________________
Dnsmasq-discuss mailing list
Dnsmasq-discuss@lists.thekelleys.org.uk
https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss

Reply via email to