7 Commits

Author	SHA1	Message	Date
Azat Khuzhin	a6d420a1c6	One cannot use /dev/null as a history file since replxx run chmod ... Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>	2023-02-20 13:59:46 +01:00
Azat Khuzhin	238d44783b	Fix flakiness of expect tests for clickhouse-client by avoiding history overlap ... Yes, all writes to the history file is done under flock, *but*, before writing the history file there is sort(), and so if you will run the following tests the 01300_client_save_history_when_terminated_long will fail: $ /src/tests/clickhouse-test --print-time -j2 01180_client_syntax_errors 01300_client_save_history_when_terminated_long 0001_select And it has nothing todo with timeouts: expect: does "" (spawn_id exp8) match glob pattern "for the history"? no f8f1dbfdaaca :) select (1, 2 expect: does "\u001b[1Gf8f1dbfdaaca :) select \u001b[0;22;33m(\u001b[0;22;32m1\u001b[0;1m,\u001b[0m \u001b[0;22;32m2\u001b[0m\u001b[J" (spawn_id exp8) match glob pattern "for the history"? no expect: does "\u001b[1Gf8f1dbfdaaca :) select \u001b[0;22;33m(\u001b[0;22;32m1\u001b[0;1m,\u001b[0m \u001b[0;22;32m2\u001b[0m\u001b[J\u001b[29G" (spawn_id exp8) match glob pattern "for the history"? no expect: timed out The "select (1, 2" is from 01180_client_syntax_errors And use real file only when the history should be preserved across runs (i.e. there are multiple invocations of clickhouse-client) CI: https://s3.amazonaws.com/clickhouse-test-reports/0/1adfbac19fed9813725d8b1df14e617b58a45d20/stateless_tests__asan__[2/2].html Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>	2023-02-20 09:51:08 +01:00
Azat Khuzhin	e4b81e6de9	Add exp_internal for expect tests ... This is to catch possible issues like 01565_reconnect_after_client_error in [1]. [1]: https://s3.amazonaws.com/clickhouse-test-reports/38417/e33f236d502fe9f7906392f4ceffaee44e3c4ce4/stateless_tests__thread__actions__[3/3].html Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>	2022-07-02 17:08:06 +03:00
Azat Khuzhin	a424b4ad56	Do not ignore eof in expect tests ... expect_after that adjusts default timeout handler, reseted eof handler, and this tell expect that it can ignore eof from read, consider the following example: <details> ```expect #!/usr/bin/expect -f exp_internal 1 log_user 1 set timeout 4 match_max 100000 expect_after { # eof { exp_continue } timeout { exit 1 } } spawn bash -c "sleep 1; echo ':) '" expect ":) " ``` ``` $ ./expect.expect < /dev/null spawn bash -c sleep 1; echo ':) ' parent: waiting for sync byte parent: telling child to go ahead parent: now unsynchronized from child spawn: returns {6614} expect: does "" (spawn_id exp4) match glob pattern ":) "? no expect: read eof ^^^ expect: set expect_out(spawn_id) "exp0" expect: set expect_out(buffer) "" ``` And with uncommented eof handler: ``` $ ./expect.expect < /dev/null spawn bash -c sleep 1; echo ':) ' parent: waiting for sync byte parent: telling child to go ahead parent: now unsynchronized from child spawn: returns {17959} expect: does "" (spawn_id exp4) match glob pattern ":) "? no expect: read eof expect: set expect_out(spawn_id) "exp0" expect: set expect_out(buffer) "" expect: continuing expect after update expect: does "" (spawn_id exp4) match glob pattern ":) "? no expect: does ":) \r\n" (spawn_id exp4) match glob pattern ":) "? yes expect: set expect_out(0,string) ":) " expect: set expect_out(spawn_id) "exp4" expect: set expect_out(buffer) ":) " ``` </details>	2021-12-29 10:02:38 +03:00
Azat Khuzhin	cb394c95d0	Enable tests that requires completion in fasttest	2021-12-17 22:14:37 +03:00
Alexey Milovidov	d604cf5573	Remove the infamous "unbundled" build	2021-11-21 13:58:26 +03:00
kssenii	bb55fb41c6	Add test	2021-10-08 07:48:05 +00:00