04.05.2020 20:57, Antoine Beaupré wrote:
> On 2020-05-01 19:10:53, Alexander Galanin wrote:
>> Package: etckeeper
>> Version: 1.18.12-1.1
>> Followup-For: Bug #884987
>>
>> The problem is still present in 1.18.12-1.
>>
>> Here is my patch for skipping mercurial's data directory:
>>
>> +++ a/pre-commit.d/20warn-problem-files 2020-05-01 18:52:57.654521579
>> +0300
>> --- b/pre-commit.d/20warn-problem-files 2020-05-01 19:05:01.103549539
>> +0300
>> @@ -10,7 +10,7 @@
>> hardlinks=$(find . -type f ! -links 1 | exclude_internal ) || true
>> elif [ "$VCS" = hg ]; then
>> special=$(find . ! -type d ! -type f ! -type l | exclude_internal) ||
>> true
>> - hardlinks=$(find . -type f ! -links 1 -exec hg status {} \; |
>> exclude_internal ) || true
>> + hardlinks=$(find . \( -name .hg -prune \) -o \( -type f ! -links 1
>> -exec hg status {} \; \) ) || true
>
> I don't get it: that's what `| exclude_internal` is designed to do, how
> is it not skipping those directories?
Error messages are coming from `hg status` which is executed _before_
exclude_internal filter. So we need to filter mercurial's internal directory
before 'hg status' call.
--
Alexander Galanin
