Hello. Now that I'm starting to work again on bringing in LedgerSMB, I wanted to put forth some ideas which seem like a good idea. This requires bringing in a lot of dependencies.
I talked about this somewhat in this thread: https://marc.info/?t=163338573000002&r=1&w=2 (I got a bad case of the flu, so I never worked on a diff) I worked on this project for the most part by myself, with some excellent help from others, but still doing almost all the actual porting work alone. I found myself extremely confused as things moved along. I submitted a good number of ports, but not even close to all of them. I knew without much warning that I had to cut loose the server I was running -current on. Too many $$ and less than a week to upload/download and transfer files to another server and home. A few hundred GB given everything else, too. I dumped the porting files into a repository which is a disastrous mess. I will start a new one that's cleaned up. At least for myself, I really, really need to have following files: Dependency_trees_for_cpanfile_modules_needing_finishing (from Perl ports generator) ports_submitted ports_submitted_but_needing_a_ping ports_submitted_but_with_issues committed_ports committed_ports_needing_to_be_updated That may seem excessive, but I have worked several times on bringing in LedgerSMB. A number of years ago, I needed 2 Perl ports committed and it was good to go. I was even about to submit some really nice code to LedgerSMB. Last try, I submitted roughly 12 ports. Getting 2 OK's wasn't possible. I had about 6 more ready to submit. Committed ports from that effort was 0. This is an observation, which might be wrong. I just don't remember any single person working on a large project showing up on the ports mailing list. Groups, yes. A few super hero porters seem to be doing some large projects that are very complex, but the road from where I am at to that level is doubtful that it will ever happen. This kind of work is hard to organize. It's hard to keep track of what has been submitted, what needs revision, what needs to be updated during the process when dependencies get updated during the project, etc. What needs to be done if multiple ports are submitted in an email, but only a few of those get committed? A new email with the missing ports to catch attention again? Does a long thread attract attention or lose it? Despite pings. So, another go at this. Better organized. Better ideas on how to submit properly. Better ideas on how to deal with certain types of ports. But I don't really want to spend weeks of work and not be able to get a second OK and thus a commit. I saw a thread that the porter had commit access and declared that getting anyone to revise his work was impossible and he would just decide on his own. I do not have any idea how well off OpenBSD is with regard to funding or how it is spent. But if there is any flexibility with funds and porters just can't get the work reviewed and committed, maybe some funding needs to be directed to help that? Bringing in something new is fun. Getting everything reviewed before I can submit it is tedious. Necessary, but sometimes it's pretty hard. When I was a little kid, when my parents were gone, I would work really hard to do the dishes and put them away. When my Mom came home, instead of praise, all I heard was that doing the dishes required wiping the counter down. My last go at this felt a lot like that. I had a lot more ready to submit once I got even a single commit. I still hate doing the dishes today. I'm very good at it, though. I would deeply appreciate it if someone could take the time to get me a second OK. I know it's a PITA to do. But it's also a PITA to get something ready to submit. -- Back to the sweat shop. ;-} Chris Bennett
