Re: [sqlite] Setting auto_vacuum=2 doesn't work after setting journal_mode=WAL

2020-02-22 Thread Wout Mertens
I do the exact same pragmas as Jens, and also in this order. I was even convinced that it was working, so I'll need to double check. The documentation might be correct if you know what to look for, but from current experience it's not obvious. Wout. On Sat., Feb. 22, 2020, 4:02 a.m. Keith

[sqlite] Consider Adding Plus Sign To Allowed Character List in althttpd

2020-02-22 Thread Carl Chave
Dr. Hipp, I'm using althttpd to serve files from a personal Linux package repository. There are 38 instances of package names which contain at least one plus sign, currently one of the characters substituted with an underscore by althttpd. This results in a 404 file not found error and makes

Re: [sqlite] unexpected sqlite_busy behaviour within transactions

2020-02-22 Thread Keith Medcalf
On Saturday, 22 February, 2020 09:26, Andy Bennett wrote: >This other process has called "BEGIN IMMEDIATE TRANSACTION" and >https://sqlite.org/rescode.html#busy says >"The BEGIN IMMEDIATE command might itself return SQLITE_BUSY, but if it >succeeds, then SQLite guarantees that no subsequent

Re: [sqlite] unexpected sqlite_busy behaviour within transactions

2020-02-22 Thread Andy Bennett
Hi, A busy_timout is honored by write commands outside of transactions and by single write commands inside transactions, but not if preceded by a read command. I did not find this behaviour in the documentation, thus it might be a bug. It's documented in the description of

Re: [sqlite] unexpected sqlite_busy behaviour within transactions

2020-02-22 Thread Igor Tandetnik
On 2/22/2020 7:50 AM, softw...@quantentunnel.de wrote: A busy_timout is honored by write commands outside of transactions and by single write commands inside transactions, but not if preceded by a read command. I did not find this behaviour in the documentation, thus it might be a bug. It's

[sqlite] unexpected sqlite_busy behaviour within transactions

2020-02-22 Thread Software
Hi A busy_timout is honored by write commands outside of transactions and by single write commands inside transactions, but not if preceded by a read command. I did not find this behaviour in the documentation, thus it might be a bug. To reproduce (in a linux terminal): sqlite3 ~/test.db