| Commit message (Collapse) | Author | Age | |
|---|---|---|---|
| * | indices can now return confidence levelsbayes | dakkar | 2020-03-16 |
| | | | | | also, tests pass again | ||
| * | actually working cache updating in ByAddresses | dakkar | 2020-03-16 |
| | | |||
| * | more async! | dakkar | 2020-03-16 |
| | | | | | | | | | | | indices now work from channels, so we have a bunch of workers doing the parsing, and one worker per index, no lock needed (the implicit lock in the react/whenever is enough) next: indices return confidence level, store returns best response next next: spamc as an index | ||
| * | always decont the arguments to pair | dakkar | 2020-03-05 |
| | | | | | | this ensures that the pair behaves like a value object (i.e. it works sensibly as a hash key) | ||
| * | nicer types | dakkar | 2020-01-26 |
| | | |||
| * | timeout when parsing a socket | dakkar | 2020-01-25 |
| | | | | | | | | | if the socket doesn't send an email message, we would be stuck waiting for it to do so since a socket is not a pipe, we can't detect "the other end has finished talking" ☹ | ||
| * | Revert "fewer temporary variables" | dakkar | 2020-01-25 |
| | | | | | | | | This reverts commit 314ee051c170c33d2c912b6750f916b73f9cb507. I think that the bug with LEAVE messes up return values inside Log::Timeline as well | ||
| * | say where things went wrong | dakkar | 2020-01-25 |
| | | |||
| * | numeric types, not-zero constant | dakkar | 2020-01-25 |
| | | |||
| * | safer socket input? | dakkar | 2020-01-25 |
| | | |||
| * | don't die if the other end closes the socket | dakkar | 2020-01-25 |
| | | |||
| * | speed-up Bayes predictions | dakkar | 2020-01-25 |
| | | | | | this is *probably* correct | ||
| * | fewer temporary variables | dakkar | 2020-01-25 |
| | | | | | | we can `return` from an inner block, even when that block is executed somewhere else: `return` is lexical! | ||
| * | nicer scan-dir signature | dakkar | 2020-01-25 |
| | | |||
| * | pre-compute more Bayes stuff | dakkar | 2020-01-11 |
| | | | | | when all my mails were indexed, it took ages to guess the mailbox | ||
| * | allow multiple maildirs on cmdline | dakkar | 2020-01-11 |
| | | |||
| * | lock when reading, too | dakkar | 2020-01-03 |
| | | |||
| * | use a set instead of an array | dakkar | 2020-01-03 |
| | | |||
| * | binding FTW | dakkar | 2020-01-03 |
| | | |||
| * | I don't understand this context/flat thing | dakkar | 2019-12-31 |
| | | |||
| * | better fix for the array type check | dakkar | 2019-12-31 |
| | | | | | | | | the slip seems to take the `:delete` as its own argument, and dies forcig explicit list context on the LHS works, although I'm not sure why it's needed | ||
| * | avoid type check failure | dakkar | 2019-12-31 |
| | | |||
| * | work around some weird newline issues | dakkar | 2019-12-31 |
| | | |||
| * | more sensible worker count default | dakkar | 2019-12-31 |
| | | |||
| * | log end-of-scan | dakkar | 2019-12-29 |
| | | |||
| * | handle receiving same file more than once | dakkar | 2019-12-29 |
| | | | | | | | `scan-dir`, because of `IO.watch`, can emit the same path multiple times; we don't want messing up the index (especially the bayesian one!) because of that | ||
| * | better scandir | dakkar | 2019-12-29 |
| | | | | | | | * breadth-first, not depth-first * emits an `End` once all files that were there at the start have been emit-ed | ||
| * | new module extension | dakkar | 2019-12-29 |
| | | |||
| * | it's called `raku` now | dakkar | 2019-12-29 |
| | | |||
| * | close files after parsing | dakkar | 2019-12-29 |
| | | |||
| * | change port from cmdline | dakkar | 2019-12-29 |
| | | |||
| * | add missing dep | dakkar | 2019-12-29 |
| | | |||
| * | better name | dakkar | 2019-12-29 |
| | | |||
| * | more standard layout & meta | dakkar | 2019-12-29 |
| | | |||
| * | kill p5 program | dakkar | 2019-12-29 |
| | | |||
| * | nicer way of finding `lib` | dakkar | 2019-12-29 |
| | | |||
| * | warn exceptions' normal stringification, not their `.perl` | dakkar | 2019-12-29 |
| | | |||
| * | 6.d has been stable a while | dakkar | 2019-12-29 |
| | | |||
| * | better predictions with small folders | dakkar | 2019-12-28 |
| | | |||
| * | no need to .key on pairs we don't care about | dakkar | 2019-12-28 |
| | | |||
| * | more types, counts seems to work better this way | dakkar | 2019-12-28 |
| | | |||
| * | one more address | dakkar | 2019-12-28 |
| | | |||
| * | forget about things when they go to 0 | dakkar | 2019-12-28 |
| | | |||
| * | naive-bayes classifying index | dakkar | 2019-12-28 |
| | | | | | half-tested by hand | ||
| * | test multi-index store | dakkar | 2019-12-28 |
| | | |||
| * | half-test for scandir | dakkar | 2019-12-28 |
| | | |||
| * | test for Store | dakkar | 2019-12-28 |
| | | |||
| * | make things faster by doing more work (?!??!) | dakkar | 2019-12-27 |
| | | |||
| * | ok, the slow bit is looking for the @separators | dakkar | 2019-12-27 |
| | | |||
| * | Log::Timeline | dakkar | 2019-12-27 |
| | | | | | apparently reading from the socket is the slowest bit | ||
 |
index : MaildirIndexer | |
| An automated system to put incoming e-mail messages in folders | dakkar |
| summaryrefslogtreecommitdiff |