The message appeared: "this message is no longer available."ġ0. I clicked the grayed out (deleted) message again.ħ. The second message displayed in the message view pane.Ħ. I clicked on the header of a second message.ĥ. The message appeared grayed out but remained in the message header pane.Ĥ. There was a refresh of the message header view pane.ģ. I deleted a mail message (highlighted it in the header pane, hit the "delete" key.Ģ. Uncannily, this just happened this morning when I got on the computer.ġ. Unfortunately it's extremely difficult to reproduce nondeterministic concurrency bugs (as you know already).! One has to examine the source code.īut here is some additional anecdotes. ![]() So, is there still something not quite right with my environment? Very possibly? But I'll post another discussion for that problem. And then I see that the message I deleted is no longer showing in the header pane.Īnd just to make the situation more complicated, I have another outstanding problem with Activity Monitor. Now I click on another mail account in the account pane, then click back to the original mail account. Previously I would just click delete two or three times impatiently. I do still notice that a message stays in the header view pane after deleting it. That being said, I am being careful to ensure that a message has downloaded from the server before even clicking on it. Oh, and by the way, after re-installing macOS from the net about a month ago, I have not had any problems with Mail. And there are many others I'm probably not thinking of. Mail should be locking any data structure that represents an index of messages (message tree, conversation tree, etc.) as well as any message manipulated. The behavior suspiciously resembles that of concurrency bugs. I suspect there is a concurrency bug somewhere in Mail. Mail should be doing locking and synchronization before accessing in-memory structures or file data. And macOS still uses POSIX threads, right.? Application code must lock a semaphore, mutex lock, read-write lock or other synchronization primitive before entering critical sections that require atomicity and one-at-a-time access. But even without file locking, there are synchronization primitives to support concurrency. When I mentioned file locking, I was thinking that Mail might lock message or meta-data files during operation. When Mail did restart finally, it re-downloaded all messages from the server and went through a fairly lengthly re-indexing cycle. Mail was not running when I deleted files and directories as the article instructs. I deleted MailData (and other sub-folders related to mail as the article instructs) only upon encountering a problem with Mail crashing on start up and core dumping. In my case I never touched any data or meta-data files while Mail was operating. Other similarly simple interactions with the Mail UI Dragging a message to move it to a "mailbox" (Apple's amusing term for sub-folder)ĥ. Double-clicking a message in the header pane to open the message in a new windowģ. Clicking a message in the message header paneĢ. I specifically saw the problem occur after doing things as simple as:ġ. When Mail works it works fine even with large messages. I try not to go higher than that a popular mail services (Google, Yahoo, etc.) limit at 50MB. But if you click the message, let it display and then delete it, I think you'll be I regularly CC myself on mail messages with photos. So sometimes Mail is consuming 45% - 80% of CPU for about 3 to 5 seconds. I noticed that the problem has not occurred since I now wait to let messages load. ![]() So, upon start up, it has some invalid pointer or invalid state of the meta-data files. ![]() I suspect that Mail does not properly "lock" the file so the index thinks the message is in your inbox but it's really not. If you try to delete the message too quickly or move it to another folder ("mailbox" in Apple's obtuse terminology) it might be causing this problem. I noticed that sometimes, a mail message appears in the message header pane but the message has not completed loading from the server (I suspect but cannot verify). I suspect there is some race condition (some concurrency bug) in the software. By the way, I have reported this to Apple many, many times.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |