ForkHi,
Current regression tests fail when block size is
NOT 8KB.
$ cat regression.diffs
diff -U3 /home/tori/pg_bigm/expected/pg_bigm.out
/home/tori/pg_bigm/results/pg_bigm.out
--- /home/tori/pg_bigm/expected/pg_bigm.out 2020-02-07
14:36:38.560021333 +0900
+++ /home/tori/pg_bigm/results/pg_bigm.out 2020-02-07
15:50:31.284920620 +0900
@@ -106,7 +106,7 @@
SELECT * FROM pg_gin_pending_stats('test_bigm_idx');
pages | tuples
-------+--------
- 43 | 249
+ 21 | 249
To cope with various block size environments,
I Attached a patch which only tests the number of
tuples rather than testing both pages and tuples.
Although this workaround makes it hard to detect
bugs in the number of pages, I don't think we
have to be so sensitive to those cases because
calculations of the number of pages are done by
PostgreSQL itself and if something goes wrong,
number of tuples will likely change.
> Fujii-san,
Thanks for your advice!
Regards,
Torikoshi Atsushi
-------------- next part --------------
テキスト形式以外の添付ファイルを保管しました...
ファイル名: dealing-with_NOT-8K_block_regress.patch
型: text/x-diff
サイズ: 2049 バイト
説明: 無し
URL: <https://lists.osdn.me/mailman/archives/pgbigm-hackers/attachments/20200213/9f0d9eb7/attachment.patch>