Den sön 24 sep. 2023 kl 12:55 skrev Yasuhito FUTATSUKI < futat...@yf.bsdclub.org>:
> Hi, > > On 2023/09/24 17:53, Daniel Sahlberg wrote: > > Den sön 24 sep. 2023 kl 07:06 skrev <futat...@apache.org>: > > > >> Author: futatuki > >> Date: Sun Sep 24 05:06:08 2023 > >> New Revision: 1912500 > >> > >> URL: http://svn.apache.org/viewvc?rev=1912500&view=rev > >> Log: > >> swig-py: Use pure Python objects as edit/parse_fns3 and their decendant > >> batons. > >> > >> Former implementation of the wrappers for svn_delta_editor_t and > >> svn_repos_parse_fns3_t C APIs had some leakage of Python reference > counts: > >> > > > > [...] > > > > I didn't review but I hope to find some time to do this later today. The > > issues fixed sounds like a very good improvement, thank you for your work > > with this! > > Thank you for reviewing it. Now I've done a series of commits related to > it, r1912500, r192501, r192502, r192503, r1912515, and r1912517. > I have reviewed these commits with only minor nits in one of them (see separate e-mail [1]). I have run the test suite with all the commits and it completes successfully. I have also run it with each of r192501-r1912517 and verified that the checks fail without r1912500. I must admit that I both Python and SWIG are relatively new to me so I don't understand this fully but I think the basic ideas are sound. I've added my vote for the backports, because of my inexperience I've only set +0 but binding fixes should be OK to backport with only one +1 and one +0. Any fixes from [1] can be added to the same backport (as I read it documentation fixes can be backported without votes). Kind regards, Daniel Sahlberg [1] https://lists.apache.org/thread/m428zfspmm1j66606mjv7ohfq9kkr1zv Kind regards, Daniel Sahlberg