On 11/19/21 at 09:09pm, Kevin Morris via pacman-dev wrote:
Signed-off-by: Kevin Morris <kevr@0cost.org> --- `alpm_unregister_all_syncdbs` errors out when a transaction is held. Since this function closes transactions as-is, moving this call below the transaction unlock step solves issues that `pyalpm` sees when it cleans up the handle.
lib/libalpm/alpm.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/lib/libalpm/alpm.c b/lib/libalpm/alpm.c index 6708e202..1c38feb8 100644 --- a/lib/libalpm/alpm.c +++ b/lib/libalpm/alpm.c @@ -107,10 +107,6 @@ int SYMEXPORT alpm_release(alpm_handle_t *myhandle) myhandle->db_local = NULL; }
- if(alpm_unregister_all_syncdbs(myhandle) == -1) { - ret = -1; - } - #ifdef HAVE_LIBCURL curl_multi_cleanup(myhandle->curlm); curl_global_cleanup(); @@ -118,6 +114,11 @@ int SYMEXPORT alpm_release(alpm_handle_t *myhandle) #endif
_alpm_handle_unlock(myhandle); + + if(alpm_unregister_all_syncdbs(myhandle) == -1) { + ret = -1; + } + _alpm_handle_free(myhandle);
return ret;
I don't see how this fixes the issue. All you've done is move alpm_unregister_all_syncdbs under _alpm_handle_unlock, but _alpm_handle_unlock doesn't clear the transaction.