björn wrote: > 2008/11/22 Ben Schmidt <[EMAIL PROTECTED]>: >>>> I am aware of bugs with Vim processes that never shut down and I will >>>> do what I can to fix these as soon as possible. >>> I suggest the appcast not be updated until we at least understand this >>> bug, and preferably not until it's fixed. Without at least knowing what >>> it is, we don't know what other consequences it might have. >>>> If you are having >>>> problems with "Vim process leaks" then I advice you to disable >>>> "quickstart" for now (and tell me if that does not take care of the >>>> problem). >> I don't have quickstart turned on, yet I saw some of the console >> messages. I haven't been able to reproduce them yet. But it seems >> quickstart just makes the problem more frequent or obvious rather than >> actually causing it. I didn't see any until the code of snapshot 37, >> though (built from source). > > There are two distinct issues here: > > 1. The one (I think) that you are seeing (and the only one I can > reproduce) which causes the following type of error message: > > com.apple.launchd[78] ([0x0-0x28028].org.vim.MacVim[592]): Stray > process with PGID equal to this dead job: PID 627 PPID 1 Vim > > This goes away if I disable the "Login shell" option so it has > something to do with forking new Vim processes. (Google this problem > and you'll find heaps of other programs causing launchd to emit this > warning.) It seems harmless, but I was hoping it would go away when > we ignored SIGCHLD.
I suspect a setsid() call (or perhaps setpgid()) would make this go away; I mentioned it in an earlier post about these kinds of issues. Let me see if I can find it... This one: http://groups.google.com/group/vim_mac/msg/ad3c5b53c0a5aa2b Perhaps try the setsid() call as discussed in there. We've tried the signal() call already, which seems to have worked as I predicted, and solved the zombie problem, but not the problem we were actually discussing. I'm not all that confident about the setsid() call, but it's something else to try; it might get rid of the messages. > 2. The more obscure bug that only happens with Quickstart enabled > which causes Vim processes to get stuck in a loop where they > continuously log the following error: > > Vim[35947] Exception caught when processing command queue: "connection > is invalid" > > I have never been able to reproduce this. In snap 38 I added a guard > that will quit the Vim process if this situation occurs to avoid > filling up people's log files. This is the bug I am most concerned > about but at least it can be avoided by disabling Quickstart (as far > as I know). It also seems to cause a new Vim process to spawn > whenever netrw is used (Issue 121). I was thinking of making a build > which logs lots of stuff and ask the people who are able to reproduce > this bug to run it and tell me what is being logged. No, Björn, on the contrary, recently I have seen the "connection is invalid" message and not the "stray process" one. I vaguely recall seeing the stray process bug months ago, but it hasn't shown up recently, I don't think. And maybe I didn't see it on my machine but only on the mailing list. I have a few thousand logged "connection is invalid" messages, though, from about three days ago, and I don't recall touching my prefs, which have quickstart disabled. There wouldn't be distinct preferences for running MacVim from the build directory, rather than from /Applications, or from a downloaded snapshot vs. one compiled from source, would there? I don't see any reason there would be, so can't think of any way quickstart could have been on. I don't recall killing any processes, either, but the log messages do end. It seems there is a frenzy of attempted communication going on for a few seconds or minutes, and then it gives up. It was happening remarkably regularly at that time, too; definitely happened a few times over the course of a couple of days. And since then, it hasn't happened again. Hmmm. Actually, I do recall something else odd happening back then, too. Something to do with mvim. I realised it was a problem due to my own stupidity, and just ignored it. But I hadn't noticed the log messages at that time. If it was about the same time as these messages were logged, maybe I have a way to reproduce it without quickstart...if I can remember what I did. I'll try. At the moment I have no idea how to reproduce the bug, even though I have seen the messages. > At any rate, none of these bugs are critical as I see it (the second > one is but you can work around it easily). I don't think they are bad > enough for me not to update the appcast (and the second bug is present > in snap 36 anyway). OK, fair enough. Ben. --~--~---------~--~----~------------~-------~--~----~ You received this message from the "vim_mac" maillist. For more information, visit http://www.vim.org/maillist.php -~----------~----~----~----~------~----~------~--~---
