Commit graph

797 commits

Author SHA1 Message Date
Scott Lystig Fritchie
0f4d5ed775 Silence dialyzer unused function clause 2015-06-25 16:36:29 +09:00
Scott Lystig Fritchie
c2faf9f499 yolo, un-do experimental type hack 2015-06-25 16:36:14 +09:00
Scott Lystig Fritchie
d9694a992a Alright, use term_to_binary() for opaque/sexp-style encoding, only 15x slower.
machi_flu1_test: timing_pb_encoding_test_... speed factor=15.12 [2.678 s] ok
2015-06-25 16:11:46 +09:00
Scott Lystig Fritchie
2763b16ca2 timing_pb_encoding_test_... speed factor=35.95 [2.730 s] ok
So, the PB style encoding of the Mpb_LL_WriteProjectionReq message
is about 35-36 times slower than using Erlang's term_to_binary()
and binary_to_term().  {sigh}
2015-06-25 16:11:44 +09:00
Scott Lystig Fritchie
5d8b648a24 All projection store protocol operations are now using Protocol Buffers!
So, there's some cheating going on, because some of the parts of
the #projection_v1{} and #p_srvr{} records aren't fully specified.
Those parts are being specified as "opaque" in the field names, e.g.

    optional bytes opaque_flap = 10;
    optional bytes opaque_inner = 11;
    required bytes opaque_dbg = 12;
    required bytes opaque_dbg2 = 13;

The serialization that's being used is erlang term sexprs.  That isn't
portable.  So if/when we really need to deal with a non-Erlang
language, we'll have to straighten this out further.
2015-06-25 15:26:35 +09:00
Scott Lystig Fritchie
841235b3b5 WIP: bugfixes, add {error, written} 2015-06-25 15:10:24 +09:00
Scott Lystig Fritchie
4fc0578a9d WIP: bugfixes, machi_flu1_test still broken 2015-06-25 15:08:40 +09:00
Scott Lystig Fritchie
d9407b76b7 WIP: dinnertime, machi_flu1_test still broken 2015-06-24 18:00:25 +09:00
Scott Lystig Fritchie
31c5bcc0c7 WIP: 1/2 of low-level projection proto finished, machi_flu1_test fails 2015-06-24 17:20:18 +09:00
Scott Lystig Fritchie
725b10ba90 Complete PB round-trip for #projection_v1{}, bleh 2015-06-24 16:13:11 +09:00
Scott Lystig Fritchie
1b0cf06f1c Fix type problem, oops 2015-06-24 14:06:17 +09:00
Scott Lystig Fritchie
2068f70700 WIP: encoding #p_srvr and #projection_v1, just starting. Damn tedious. 2015-06-24 12:50:37 +09:00
Scott Lystig Fritchie
817efb2b15 machi_pb_high_client: always be checksumming 2015-06-23 17:37:47 +09:00
Scott Lystig Fritchie
22ae33e298 Merge branch 'slf/pb-api-experiment1' 2015-06-23 17:26:55 +09:00
Scott Lystig Fritchie
d3b0b7fdc5 Clean up some dialyzer complaints 2015-06-23 17:26:15 +09:00
Scott Lystig Fritchie
727b2a987d ROTFL forgot to add src/machi_pb_server.erl 2015-06-23 17:22:45 +09:00
Scott Lystig Fritchie
ceebe3d491 WIP: list_files #2 2015-06-23 17:17:14 +09:00
Scott Lystig Fritchie
73f71c406e WIP: list_files end-to-end! 2015-06-23 17:08:15 +09:00
Scott Lystig Fritchie
6722b3c0f1 WIP: checksum_list incomplete implementation.... 2015-06-23 16:53:06 +09:00
Scott Lystig Fritchie
6e77a4ea74 WIP: read_chunk end-to-end! 2015-06-23 16:24:08 +09:00
Scott Lystig Fritchie
44c22bf752 WIP: read_chunk #1 2015-06-23 15:34:48 +09:00
Scott Lystig Fritchie
a8782eed5a WIP: write_chunk #1 2015-06-23 15:13:13 +09:00
Scott Lystig Fritchie
cb06c53dc0 WIP: PB append_chunk end-to-end works! 2015-06-23 14:45:24 +09:00
Scott Lystig Fritchie
5ef499ec73 WIP: append_chunk #1 2015-06-23 14:08:10 +09:00
Scott Lystig Fritchie
bb8e725c26 WIP: 'auth' request placeholders 2015-06-22 18:16:15 +09:00
Scott Lystig Fritchie
db7f1476b9 WIP: 'echo' request works end-to-end, yay! 2015-06-22 18:04:17 +09:00
Scott Lystig Fritchie
3d05f543df WIP: new test case is failing, quick fix soon 2015-06-22 17:49:07 +09:00
Scott Lystig Fritchie
70616d3bd5 The FAQ grows, part 2e 2015-06-22 15:11:31 +09:00
Scott Lystig Fritchie
25e06da4b5 The FAQ grows, part 2d 2015-06-22 00:46:47 +09:00
Scott Lystig Fritchie
8cfc25db36 The FAQ grows, part 2c 2015-06-22 00:38:16 +09:00
Scott Lystig Fritchie
f55897afcb The FAQ grows, part 2b 2015-06-22 00:31:41 +09:00
Scott Lystig Fritchie
d0c39a7ec1 The FAQ grows, part 2 2015-06-22 00:26:25 +09:00
Scott Lystig Fritchie
48e4bf2c1a The FAQ grows 2015-06-22 00:09:35 +09:00
Scott Lystig Fritchie
e0fd2e909c Typo 2015-06-21 17:54:17 +09:00
Scott Lystig Fritchie
73b6a90e78 Create FAQ.md 2015-06-21 17:47:07 +09:00
Scott Lystig Fritchie
b65293f391 Merge branch 'slf/pb-api-experiment1' 2015-06-21 15:10:54 +09:00
Scott Lystig Fritchie
dc9f272c44 Nearly dumbest-possible Protocol Buffers client request & response round trip 2015-06-19 17:21:04 +09:00
Scott Lystig Fritchie
c4bdeee4da Oops, add missing src/machi_dt.erl 2015-06-19 17:20:49 +09:00
Scott Lystig Fritchie
0cdaee32f8 Egadz, edoc doesn't use preprocessor {sigh} 2015-06-19 16:24:57 +09:00
Scott Lystig Fritchie
984b4f7a86 Dialyzer tightening and subsequent cleanup 2015-06-19 16:04:34 +09:00
Scott Lystig Fritchie
1372bd9594 {sigh} add filter-dialyzer-dep-warnings 2015-06-19 15:22:07 +09:00
Scott Lystig Fritchie
3c300bb9f1 Add write_chunk() to machi_cr_client.erl 2015-06-19 14:49:09 +09:00
Scott Lystig Fritchie
40c0a72b48 Add test/machi_pb_test.erl, finish PB refactoring 2015-06-19 13:00:28 +09:00
Scott Lystig Fritchie
a82bd68f3c Overhaul the 0.1 PB definition. Again.
Many thanks to @seancribbs for a suggestion to avoid the PB design
mistake/feature of the original Riak KV PB API.
2015-06-19 12:28:31 +09:00
Scott Lystig Fritchie
87b636a349 WIP: PB wrestling 2015-06-18 17:31:48 +09:00
Scott Lystig Fritchie
1f026aee30 Change the name-game-sketch.org file naming scheme slightly
Patch the current draft, which does not include storing the
cluster-of-clusters placement key K in the file name.  I've now convinced
myself that K must be present in the file name to allow file migration to
work correctly as described.
2015-06-18 16:21:54 +09:00
Scott Lystig Fritchie
37a8c1c124 WIP: PB wrestling 2015-06-18 16:16:23 +09:00
Scott Lystig Fritchie
e5673b5e20 First attempt at Protocol Buffers .proto + infrastructure 2015-06-17 16:12:20 +09:00
Scott Lystig Fritchie
3ce3fb93b9 Use infinity timeout for sanity check 2015-06-17 12:42:53 +09:00
Scott Lystig Fritchie
e7e5ab6f87 ROTFL, time is hard. The current year is *2015*. 2015-06-17 12:30:48 +09:00