owner stringclasses 14
values | repo stringclasses 14
values | id int64 116k 4.61B | issue_number int32 1 181k | author stringlengths 1 39 | body stringlengths 1 262k | created_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:16:18 | updated_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:22:56 | reactions unknown | author_association stringclasses 5
values |
|---|---|---|---|---|---|---|---|---|---|
ClickHouse | ClickHouse | 542,326,645 | 7,339 | amosbird | > Does not compile for me:
Yeah, you need to grab double-conversion ( can just use the contrib one)
> What exactly numbers are these?
Don't remember now, but it's easy to reproduce with the given code snippet. | 2019-10-15T17:42:43 | 2019-10-15T17:42:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 541,345,355 | 7,288 | richardartoul | > > If you're ok with this feature and it's not too much work I could try and implement it myself also if you could point me in the right direction.
>
> Yes, I'm totally Ok with that.
>
> > After looking through the code it seems like I would just need to add an optional call to disableChecksumming in Connection... | 2019-10-12T17:29:35 | 2019-10-12T17:29:35 | {} | NONE |
ClickHouse | ClickHouse | 541,423,188 | 7,296 | blinkov | I don't see any reason why do we need it there. You're welcome though! | 2019-10-13T14:31:23 | 2019-10-13T14:31:23 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,510,795 | 7,300 | alexey-milovidov | > gcc bugs
So, how did you build it? With newer version of gcc or with clang? | 2019-10-14T05:52:06 | 2019-10-14T05:52:06 | {} | MEMBER |
ClickHouse | ClickHouse | 541,511,232 | 7,300 | amosbird | > So, how did you build it?
by blacklisting those functions and disabling unneeded features (capnp). I've already tested with gcc 8.3, 9.1, 9.2. I tried clang (ubsan build) without reverting the source changes. I'll try a full build with clang. | 2019-10-14T05:54:36 | 2019-10-14T05:54:36 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 541,571,926 | 7,301 | alexey-milovidov | Does it reproduce on recent releases? | 2019-10-14T09:18:40 | 2019-10-14T09:18:40 | {} | MEMBER |
ClickHouse | ClickHouse | 541,650,339 | 7,267 | amosbird | > Probably we have to try clang-format again, so that we don't have to worry about formatting anymore.
Yeah, that's very wanted.
> To complete this PR, we should convert find() to the new interface and remove the old interface altogether.
Hmm, but it's already using the new interface now. `lookupResultGetKe... | 2019-10-14T12:32:15 | 2019-10-14T12:32:15 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 541,747,606 | 7,314 | den-crane | Reproducible example:
create table ax (A Int64, B Int64) Engine=Memory;
create table bx (A Int64) Engine=Memory;
select * from bx, ax where ax.A=bx.A and ax.B in (1,2);
Received exception from server (version 19.16.1):
Code: 403. DB::Exception: Received from localhost:9000. DB::Exception: Invalid expressio... | 2019-10-14T15:28:57 | 2019-10-14T15:28:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,808,437 | 7,326 | drstrangelooker | I think I was using an older version of the server when I noted this. I'll close. | 2019-10-14T17:23:59 | 2019-10-14T17:23:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,048,743 | 7,324 | alexey-milovidov | @hczhcz Yes. And for rand() we should have a setting controlling its behaviour for backward compatibility. | 2019-10-15T05:51:48 | 2019-10-15T05:51:48 | {} | MEMBER |
ClickHouse | ClickHouse | 542,276,075 | 7,336 | akuzm | What do other DBs do?
Postgres returns null:
```
# create table ttt (a int);
# select avg(a), max(a) from ttt;
avg | max
--------+--------
<null> | <null>
(1 row)
```
Our other functions like max() return the default value for type (that is 0), if remember correctly.
Overall, we should strive to h... | 2019-10-15T15:40:33 | 2019-10-15T15:40:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,380,925 | 7,019 | silviucpp | Only problem I can see is with CREATING TEMPORARY TABLES.
Based on docs: "Temporary tables disappear when the session ends, including if the connection is lost."
But if I run a query like `CREATE TEMPORARY TABLE `temp1` AS SELECT number FROM system.numbers FORMAT JSON` no matter If I close the connection the quer... | 2019-10-15T20:02:55 | 2019-10-15T20:02:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,390,484 | 7,334 | filimonov | Workaround:
```
SELECT * FROM test2 WHERE positionUTF8(test, 'ß') != 0 or positionUTF8(test, 'SS') != 0
```
BTW: https://en.wikipedia.org/wiki/Capital_%E1%BA%9E
More general: https://github.com/ClickHouse/ClickHouse/issues/7341 | 2019-10-15T20:28:40 | 2019-10-15T20:46:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,497,309 | 5,883 | misazsq | I miss the same question. But when i try to execute the sql in my logs of the task, I found some mistake。Like this：
DB::Exception: No macro cluster in config
Because I use layer but not cluster in my config.xml | 2019-10-16T03:38:27 | 2019-10-16T03:38:27 | {} | NONE |
ClickHouse | ClickHouse | 541,430,695 | 7,298 | alexey-milovidov | ```
/// Count number of calls where all arguments are not NULL.
class AggregateFunctionCountNotNullVariadic ...
```
You can remove this code. | 2019-10-13T15:55:14 | 2019-10-13T15:55:14 | {} | MEMBER |
ClickHouse | ClickHouse | 541,743,645 | 7,314 | nhudinh2103 | Thanks for quick response, but I currently use clickhouse-jdbc with Mondrian schema and it generates sql like above
```
select *
from `web_olap`.`date_dim` as `date_dim`, `web_olap`.`sale_fact` as `sale_fact`
where `sale_fact`.`date_dim_id` = `date_dim`.`id`
and (`sale_fact`.`order_id` in (1, 2))
```
Can c... | 2019-10-14T15:21:48 | 2019-10-14T15:21:48 | {} | NONE |
ClickHouse | ClickHouse | 541,840,610 | 7,293 | vdimir | > It does not build with clang, but it's very easy to fix.
Thanks, now I see issue with `UNUSED`. I’ll fix it in coming days | 2019-10-14T18:24:45 | 2019-10-14T18:24:45 | {} | MEMBER |
ClickHouse | ClickHouse | 542,078,219 | 7,274 | clemrodriguez | Hi,
This PR has a bug, that I need to fix. Could you, please wait before reviewing it.
Regards. | 2019-10-15T07:27:55 | 2019-10-15T07:27:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,106,027 | 7,285 | blinkov | @bi122 those extra spaces were not removed | 2019-10-15T08:43:09 | 2019-10-15T08:43:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,274,897 | 7,318 | drstrangelooker | Of course no SQL dialect is 100% compatible with another.
Keep in mind that these inconsistencies are barriers to adoption. For a
well established platform, those barriers help keep users on their own
platform. For the new upstart, however, the barriers keep the newer product
from getting a critical mass of users.
On... | 2019-10-15T15:37:56 | 2019-10-15T15:37:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,276,705 | 7,336 | akuzm | If you have the related issues at hand, maybe it makes sense to link them all here -- I remember seeing a couple of them. | 2019-10-15T15:41:55 | 2019-10-15T15:41:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,508,785 | 7,328 | hczhcz | It is also possible to implement `arrayCompact` on other comparable types (strings, tuples, ...). | 2019-10-16T04:33:03 | 2019-10-16T04:33:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,510,657 | 7,300 | alexey-milovidov | Related: https://github.com/ClickHouse/ClickHouse/issues/7302 | 2019-10-14T05:51:27 | 2019-10-14T05:51:27 | {} | MEMBER |
ClickHouse | ClickHouse | 541,526,615 | 7,300 | amosbird | Clang-9 builds fine, however, linking statically with external icu fails, same in https://github.com/ClickHouse/ClickHouse/issues/6460#issuecomment-536404533 | 2019-10-14T07:09:47 | 2019-10-14T07:09:47 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 541,641,176 | 7,267 | akuzm | To complete this PR, we should convert find() to the new interface and remove the old interface altogether. | 2019-10-14T12:12:55 | 2019-10-14T12:12:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,770,791 | 7,316 | den-crane | Duplicate https://github.com/ClickHouse/ClickHouse/issues/4860
WIP https://github.com/ClickHouse/ClickHouse/pull/7170
| 2019-10-14T16:10:19 | 2019-10-14T16:10:19 | {
"eyes": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,116,138 | 6,897 | slimtom95 | Thanks for the detailed explanation! | 2019-10-15T09:08:16 | 2019-10-15T09:08:16 | {} | NONE |
ClickHouse | ClickHouse | 542,572,410 | 7,333 | godfreyd | > Potato is internal Yandex technology. So, you must provide complete explanation about what it is, what is does, and how any external contributor should treat this config - directly by comments in YAML file (I hope YAML has support for comments).
I will. If YML does not support comments, I can explain in the README... | 2019-10-16T07:52:07 | 2019-10-16T07:52:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,684,404 | 7,337 | filimonov | Some minimal example of data & query used in MV will be helpful. | 2019-10-16T12:51:18 | 2019-10-16T12:51:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,704,226 | 7,337 | igor-sh8 | Hi.
Destination table create query:
```
CREATE TABLE test.TestTableTable (`Date` Date, `Id` Int32, `Name` String, `ArrayStringId` String, `ArrayIntId` Int32) ENGINE = MergeTree() PARTITION BY Date ORDER BY Id SETTINGS index_granularity = 8192
```
MV create query:
```
CREATE MATERIALIZED VIEW test.TestTableView T... | 2019-10-16T13:36:15 | 2019-10-16T13:36:15 | {} | NONE |
ClickHouse | ClickHouse | 541,723,617 | 7,204 | jatin33 | Hi @alexey-milovidov,
Can I work on this?
Please assign it to me if possible.
Thanks in advance.
| 2019-10-14T14:48:20 | 2019-10-14T14:48:20 | {} | NONE |
ClickHouse | ClickHouse | 541,776,434 | 3,214 | drstrangelooker | Observed with version 19.15.2.2 as well. | 2019-10-14T16:20:55 | 2019-10-14T16:20:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,836,922 | 7,319 | drstrangelooker | Thanks. I was able to implement that for YYYYMMDD. | 2019-10-14T18:18:06 | 2019-10-14T18:18:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,315,452 | 7,338 | den-crane | создавать новую таблицу с ORDER BY (a, b, с)
и переливать данные в нее | 2019-10-15T17:14:22 | 2019-10-15T17:14:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,432,628 | 7,338 | den-crane | если в индексе будут тысячи столбцов, то работать это не будет, даже на сотне столбцов индекс будет занимать слишком много памяти (индекс целиком помещается в память).
Более того на практике уже после 4,5 столбцов, последующие столбцы обычно не изменяют порядок строк.
В общем это бессмыслица.
И нет никаких плано... | 2019-10-15T22:32:01 | 2019-10-15T22:34:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,349,054 | 7,288 | alexey-milovidov | > When you say reading from tables, that only includes local tables right?
Yes.
> As far as I can tell the remote table feature will use a client connection to stream blocks in so disabling checksumming there should stop it from checksumming data coming back from remote tables.
True. This data will be first re... | 2019-10-12T18:14:46 | 2019-10-12T18:14:46 | {} | MEMBER |
ClickHouse | ClickHouse | 541,531,837 | 6,943 | cyfdecyf | @alexey-milovidov @filimonov We limit query memory limit to 4GB and encounter memory limit exception when inserting into tables with more than 2000 columns. Our current workaround is to create multiple tables to reduce columns per table.
It would be nice to reduce pre-allocated column buffer size when number of colu... | 2019-10-14T07:31:21 | 2019-10-14T07:31:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,571,156 | 6,460 | alexey-milovidov | BTW, it's Ok to "steal" this task from @abyss7 | 2019-10-14T09:16:54 | 2019-10-14T09:16:54 | {} | MEMBER |
ClickHouse | ClickHouse | 541,802,516 | 7,293 | alexey-milovidov | It does not build with clang, but it's very easy to fix. | 2019-10-14T17:12:41 | 2019-10-14T17:12:41 | {} | MEMBER |
ClickHouse | ClickHouse | 542,155,352 | 7,303 | 4ertus2 | Related #7214 | 2019-10-15T10:53:34 | 2019-10-15T10:53:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,167,425 | 7,331 | hczhcz | See also #6741 | 2019-10-15T11:30:56 | 2019-10-15T11:30:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,660,224 | 7,350 | akuzm | Good catch, although I'm not sure that these changes would be sufficient to prevent future mistakes of this kind. Come to think of it, capturing the thread's stack to use it in another threads is perverse, but we don't have other way to represent return values with this API. Even if we had it, there's no way to prevent... | 2019-10-16T11:42:25 | 2019-10-16T11:42:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,418,882 | 7,089 | danlark1 | #7299 | 2019-10-13T13:32:18 | 2019-10-13T13:32:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,606,021 | 7,267 | akuzm | As before, it's hard to see the significant changes among all the formatting changes. Probably we have to try clang-format again, so that we don't have to worry about formatting anymore. | 2019-10-14T10:43:59 | 2019-10-14T10:43:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,619,911 | 7,275 | PhantomPhreak | Aha, now it's clear, thank you.
> Maybe we should introduce an option to reset flag on later levels? Smth like that:
It looks like a way to override a `<readonly/>` setting from a **default** profile. But what if i'd like to set some min/max values, not just to make it changeable again? Like in my example, there ... | 2019-10-14T11:19:14 | 2019-10-14T11:19:14 | {} | NONE |
ClickHouse | ClickHouse | 542,298,436 | 7,186 | akuzm | The performance test not finishing is related to problems in master. This PR shouldn't have performance consequences, since it's for a new type of join. Will merge as is. | 2019-10-15T16:31:02 | 2019-10-15T16:31:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,690,749 | 7,179 | akuzm | Can't reproduce the ubsan failure locally even on the deb package from the CI. Rebased to master hoping that it'll go away. | 2019-10-14T13:52:44 | 2019-10-14T13:52:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,787,113 | 7,318 | den-crane | as documented https://clickhouse.yandex/docs/en/query_language/functions/#strong-typing
Strong typing¶
In contrast to standard SQL, ClickHouse has strong typing. In other words, it doesn't make implicit conversions between types. Each function works for a specific set of types. This means that sometimes you need ... | 2019-10-14T16:41:56 | 2019-10-14T16:41:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,806,513 | 7,326 | den-crane | https://clickhouse.yandex/docs/en/query_language/functions/date_time_functions/#tostartofweek-t-mode
```
SELECT toStartOfWeek(today())
┌─toStartOfWeek(today())─┐
│ 2019-10-13 │
└────────────────────────┘
```
------
mode=1 -> Monday
```
SELECT toStartOfWeek(today(), 1)
┌─toStartOfWeek(today(),... | 2019-10-14T17:20:21 | 2019-10-14T17:22:15 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,995,824 | 5,295 | isaachui | This seems to be happening to our servers too. Not entirely sure what the cause is. It's also quite hard to reproduce this error. | 2019-10-15T01:29:23 | 2019-10-15T01:29:23 | {} | NONE |
ClickHouse | ClickHouse | 542,199,767 | 7,134 | proller | ```
[1/12] Linking CXX executable clickhouse_build/dbms/src/Interpreters/tests/select_query
FAILED: clickhouse_build/dbms/src/Interpreters/tests/select_query
: && /usr/bin/ccache /usr/bin/c++ -fdiagnostics-color=always -fsized-deallocation -pipe -msse4.1 -msse4.2 -mpopcnt -fno-omit-frame-pointer -Wall -Wnon-vi... | 2019-10-15T13:01:58 | 2019-10-15T13:01:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,271,355 | 7,318 | filimonov | > In most databases
We want ClickHouse to be better than most of databases - at least from performance perspective ;) That leads to some compromises, and ClickHouse SQL dialect is not 100% consistant with SQL standards.
But we collect information about all inconsistancies and maybe one day we'll be able to addres... | 2019-10-15T15:30:38 | 2019-10-15T15:30:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,375,174 | 7,019 | silviucpp | For me on master works ok. When I close the connection the query is canceled and I can see in logs:
```
DB::Exception: Query was cancelled (version 19.15.1.1)
```
BTW if you send the requests via POST you need to specify readonly=2 as well in the query params.
| 2019-10-15T19:47:21 | 2019-10-15T19:47:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,344,281 | 7,281 | alexey-milovidov | Could you please send a file that reproduces this issue?
(you can post it here or send it to clickhouse-feedback@yandex-team.com) | 2019-10-12T17:16:34 | 2019-10-12T17:16:34 | {} | MEMBER |
ClickHouse | ClickHouse | 541,558,141 | 7,303 | nhudinh2103 | My clickhouse version is **19.15.2.2** | 2019-10-14T08:46:40 | 2019-10-14T08:46:40 | {} | NONE |
ClickHouse | ClickHouse | 542,207,045 | 7,330 | den-crane | Null in CSV should be encoded as \N
Or use input_format_null_as_default
```
create table XDate(A Nullable(Date), O String) Engine = Memory;
insert into XDate format CSV \N,aaaa
SELECT * FROM XDate
┌────A─┬─O────┐
│ ᴺᵁᴸᴸ │ aaaa │
└──────┴──────┘
set input_format_null_as_default = 1
insert into XDate fo... | 2019-10-15T13:18:26 | 2019-10-15T13:18:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,569,213 | 7,345 | alex-krash | @KochetovNicolai , I may implement loss functions (we are currently using them for timeseries analysis already, and I was thinking that they should be implemented as part of CH). Should I create a separate issue for discussion of it? | 2019-10-16T07:43:56 | 2019-10-16T07:43:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,639,467 | 7,301 | DNEGREEV | It's works in version 19.15.3.6
Thank you! | 2019-10-16T10:37:41 | 2019-10-16T10:37:41 | {} | NONE |
ClickHouse | ClickHouse | 542,644,231 | 7,338 | PlatonB | > последующие столбцы обычно не изменяют порядок строк.
Странно. В материале https://ruhighload.com/Колоночные+базы+данных написано "все данные в колоночной базе данных обычно хранятся в отсортированном виде (**каждая колонка отдельно**)". Насколько я понял по примечанию в скобках, сортировка столбца _a_ не должна в... | 2019-10-16T10:52:22 | 2019-10-16T10:52:22 | {} | NONE |
ClickHouse | ClickHouse | 541,146,998 | 7,278 | nvartolomei | ~~This PR fixes the issue if `skip_unavailable_shards` is set on the connection level or session level.~~
Edit: Last commit fixes all cases. | 2019-10-11T17:08:23 | 2019-10-13T14:31:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,751,285 | 7,204 | alexey-milovidov | Note that GitHub does not allow to assign issues to people outside of the teams.
https://github.com/isaacs/github/issues/100 | 2019-10-14T15:35:27 | 2019-10-14T15:35:35 | {} | MEMBER |
ClickHouse | ClickHouse | 542,246,366 | 7,186 | akuzm | Overall looks good to me. | 2019-10-15T14:39:32 | 2019-10-15T14:39:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,527,904 | 7,344 | yd1261137568 | Execute the command 'show tables' on the client, and the server reports an error | 2019-10-16T05:44:49 | 2019-10-16T05:44:49 | {} | NONE |
ClickHouse | ClickHouse | 542,685,106 | 7,348 | filimonov | Actually it's not an error, it's just a trace info.
Nevertheless it is confusing | 2019-10-16T12:53:04 | 2019-10-16T12:53:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,577,513 | 7,294 | alexey-milovidov | > Functional stateless tests (release) — Some queries hung
- unrelated. @Akazz | 2019-10-14T09:32:03 | 2019-10-14T09:32:03 | {} | MEMBER |
ClickHouse | ClickHouse | 541,597,372 | 7,254 | stavrolia | Continuation in #7307 | 2019-10-14T10:19:59 | 2019-10-14T10:19:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,776,417 | 7,297 | hczhcz | Is it because of [Hacktoberfest](https://hacktoberfest.digitalocean.com)? | 2019-10-14T16:20:52 | 2019-10-14T16:20:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,790,700 | 7,319 | den-crane | ```
SELECT parseDateTimeBestEffort('20190101')
┌─parseDateTimeBestEffort('20190101')─┐
│ 2019-01-01 00:00:00 │
└─────────────────────────────────────┘
``` | 2019-10-14T16:49:17 | 2019-10-14T16:49:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,021,771 | 7,324 | hczhcz | Probably it is better to have `randInt32()`, `randInt64()`, `randFloat32()`, `randFloat64()`, etc.? | 2019-10-15T03:37:48 | 2019-10-15T03:37:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,427,884 | 7,338 | PlatonB | Будет ли оптимизация на шаге 3 столь же маленькой, если столбцов - тысячи, а досортировать надо один? | 2019-10-15T22:15:08 | 2019-10-15T22:15:08 | {} | NONE |
ClickHouse | ClickHouse | 541,579,137 | 7,283 | alexey-milovidov | Please update the tests. | 2019-10-14T09:36:30 | 2019-10-14T09:36:30 | {} | MEMBER |
ClickHouse | ClickHouse | 541,749,530 | 7,204 | alexey-milovidov | @jatin33 Yes, you can work on this task. | 2019-10-14T15:32:22 | 2019-10-14T15:32:22 | {} | MEMBER |
ClickHouse | ClickHouse | 542,054,680 | 7,325 | hczhcz | The weekday names are language-dependent. It is better to come with a setting item. | 2019-10-15T06:14:15 | 2019-10-15T06:14:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,284,877 | 7,336 | den-crane | ```
create table tttx (a Nullable(Int64)) engine=Memory;
select avg(a), max(a) from tttx;
┌─avg(a)─┬─max(a)─┐
│ ᴺᵁᴸᴸ │ ᴺᵁᴸᴸ │
└────────┴────────┘
```
I think it should be 0 for non-nullable types, not the NaN.
1. It's more natural conversion null -> 0
```
SELECT assumeNotNull(CAST(NULL, 'Nullable(I... | 2019-10-15T15:59:29 | 2019-10-15T16:02:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,387,399 | 7,338 | filimonov | Данные уже отсортированы по `(a,b)` но не по `c`.
И в таком виде записаны на диск. Верно?
Чтобы они стали отсортированы по (a,b,c) - нужно их:
1) прочитать с диска
2) распаковать
3) пересортировать (в этом месте возможна некая отпимизация с учетом того что данные уже посортированы по a,b)
4) запаковать
5) за... | 2019-10-15T20:20:23 | 2019-10-15T20:23:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,588,131 | 7,347 | filimonov | Wierd...
```
SELECT
test,
toTypeName(test),
IF(test = 0, 1, 0)
FROM default.test
┌─test─┬─toTypeName(test)──┬─IF(equals(test, 0), 1, 0)─┐
│ ᴺᵁᴸᴸ │ Nullable(Float64) │ 1 │
└──────┴───────────────────┴───────────────────────────┘
WITH materialize(CAST(NULL, 'Nul... | 2019-10-16T08:29:18 | 2019-10-16T08:29:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,394,310 | 7,281 | thyn | I can't reproduce it. If I retry it is inserted perfectly. I'm using c# as client using http interface. I've changed format from deflate to gzip on one node and problem disappeared. I know that gzip and deflate different in format. It' s strange. But everything is ok for 2 days. Other node using deflate codec still ha... | 2019-10-13T07:27:53 | 2019-10-13T07:27:53 | {} | NONE |
ClickHouse | ClickHouse | 541,473,891 | 7,300 | amosbird | Some findings when building and testing arm build.
- ### gcc bugs
1. capnp fails to compile
2. bit rotation
```
during RTL pass: expand
In file included from /data1/ClickHouse/src/dbms/src/Functions/bitRotateLeft.cpp:2:
/data1/ClickHouse/src/dbms/src/Functions/FunctionBinaryArithmetic.h: In static member ... | 2019-10-14T00:22:58 | 2019-10-14T00:23:23 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 541,556,937 | 7,279 | alesapin | Reproduces on current master :( | 2019-10-14T08:43:13 | 2019-10-14T08:43:13 | {} | MEMBER |
ClickHouse | ClickHouse | 541,591,607 | 7,057 | zhang2014 | Performance test failures are not related to modifications and I think this pr is ready for review. | 2019-10-14T10:06:48 | 2019-10-14T10:06:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,750,031 | 7,204 | alexey-milovidov | I can merge a pull request from @dsmith47, @divc99, @jatin33 - whoever will do it first or do it better. | 2019-10-14T15:33:19 | 2019-10-14T15:33:19 | {} | MEMBER |
ClickHouse | ClickHouse | 542,298,775 | 7,134 | amosbird | @proller Hmm, what build variant did you use to produce that link error? I tried CI release and unitest builds, both compile fine in current master. That error means `AsynchronousMetrics.cpp` need to be pulled into target `clickhouse_new_delete` which is the sole place to link malloc lib. | 2019-10-15T16:31:52 | 2019-10-15T16:31:52 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 542,324,976 | 7,331 | akuzm | Looks very useful. Please add a mention of these combinators to the docs (it would be in docs/en/query_language/agg_functions/combinators.md).
A recent related issue: https://github.com/ClickHouse/ClickHouse/issues/7336
| 2019-10-15T17:38:33 | 2019-10-15T17:38:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,556,683 | 7,344 | alexey-milovidov | What version of ClickHouse do you use? Did you build your own fork of ClickHouse?
Hint: divizion by zero in std::unordered_map in gcc's standard library happens if someone was clobbered FPU state (e.g. using MMX instruction set). | 2019-10-16T07:10:13 | 2019-10-16T07:10:13 | {} | MEMBER |
ClickHouse | ClickHouse | 541,621,058 | 7,305 | pachico | OK, found the issue.
system_profile is default, which is readonly.
Log was showing it but when buffer was created, now when trying to flush, which is why I didn't see it.
My fault. | 2019-10-14T11:21:50 | 2019-10-14T11:21:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,719,870 | 7,292 | akuzm | Still seeing a regression on the synthetic emplace test (string_hash_map.cpp, the measured value is seconds elapsed to insert 50000000 elements of term48):
```
┏━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━┳━━━━━━┳━━━━━━┳━━━━━━┳━━━━━━━━━┓
┃ version ┃ count() ┃ min ┃ med ┃ max ┃ q95-q05 ┃
┡━━━━━━━━━━━━━━━━━━━━━... | 2019-10-14T14:41:35 | 2019-10-14T14:44:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,353,670 | 7,338 | PlatonB | В том-то и суть, что нужно без перезаливов. Возможно, есть метод, который забыли отразить в документации. | 2019-10-15T18:50:46 | 2019-10-15T18:50:46 | {} | NONE |
ClickHouse | ClickHouse | 542,559,445 | 7,333 | alexey-milovidov | Potato is internal Yandex technology. So, you must provide complete explanation about what it is, what is does, and how any external contributor should treat this config - directly by comments in YAML file (I hope YAML has support for comments). | 2019-10-16T07:17:56 | 2019-10-16T07:18:11 | {} | MEMBER |
ClickHouse | ClickHouse | 541,615,257 | 7,278 | nvartolomei | @alexey-milovidov looks like we might need lock on settings object or cloning for settings? I was trying to lift applying settings before we insert query in process list, but it will break settings on sub-queries. | 2019-10-14T11:07:28 | 2019-10-14T11:07:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,711,745 | 7,313 | den-crane | tuple(cast(ruleId,'String')) or upgrade CH
---------------
```
Connected to ClickHouse server version 18.14.18 revision 54409.
:) select tuple( x) from (select toLowCardinality('x') x)
DB::Exception: DataTypeLowCardinality is supported only for numbers, strings, Date or DateTime, but got Tuple(String)
:)... | 2019-10-14T14:28:17 | 2019-10-14T14:32:59 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,730,774 | 7,303 | den-crane | >INNER JOIN web_olap.product_dim product_dim ON sale_fact.product_dim_id = product_dim.id
INNER JOIN web_olap.product_dim **product_dim1** ON sale_fact.product_dim_id = **product_dim1**.id | 2019-10-14T15:00:03 | 2019-10-14T15:00:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,739,381 | 7,314 | den-crane | it should be
```
select *
from `web_olap`.`date_dim` as `date_dim`
join `web_olap`.`sale_fact` as `sale_fact` on `sale_fact`.`date_dim_id` = `date_dim`.`id`
where (`sale_fact`.`order_id` in (1, 2))
```
and for better performance
```
select *
from `web_olap`.`date_dim` as `date_dim`
... | 2019-10-14T15:14:30 | 2019-10-14T15:16:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,756,307 | 6,161 | akuzm | Something new in FunctionArrayMapped on stateful tests:
https://clickhouse-test-reports.s3.yandex.net/0/f0409348a1daa04a79e898a0db68147db4425077/functional_stateful_tests_(memory).html | 2019-10-14T15:44:14 | 2019-10-14T15:44:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 541,806,338 | 7,326 | alexey-milovidov | It is named `toMonday` intentionally to avoid confusion between europeian / north american / muslim weeks. | 2019-10-14T17:20:00 | 2019-10-14T17:20:00 | {} | MEMBER |
ClickHouse | ClickHouse | 541,866,283 | 6,229 | richardartoul | Is there any documentation on how supertypes work? What are they currently used for? My understanding was that they exist only to resolve situations like this so I'm confused as to why this would not be supported.
If these columns were part of the same table would it work then? | 2019-10-14T19:13:55 | 2019-10-14T19:13:55 | {} | NONE |
ClickHouse | ClickHouse | 542,320,446 | 7,339 | akuzm | > https://wandbox.org/permlink/2qhQc2jXepz8mbDj illustrates why it's needed.
Does not compile for me:
```
prog.cc:6:10: fatal error: double-conversion/double-conversion.h: No such file or directory
6 | #include <double-conversion/double-conversion.h>
```
Is the idea that musl exp10 gives wrong result for ... | 2019-10-15T17:26:57 | 2019-10-15T17:26:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,380,891 | 7,223 | filimonov | > when i run a query it takes 11 second on both tables, but now (after one month) its take 23 second on both tables, i didn't do any change in both tables !!
Are you sure it's not a page cache issue? Is that result stable when you repeat the queries multiple times? Maybe merged parts are too big for the page cache?
... | 2019-10-15T20:02:51 | 2019-10-15T20:06:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,467,530 | 7,342 | den-crane | Timezone is a attribute of a DateTime type.
```
SELECT
toDateTime(now(), 'Europe/Moscow') AS column,
toTypeName(column) AS x
┌──────────────column─┬─x─────────────────────────┐
│ 2019-10-16 04:12:04 │ DateTime('Europe/Moscow') │
└─────────────────────┴───────────────────────────┘
``` | 2019-10-16T01:12:30 | 2019-10-16T01:12:30 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 542,645,401 | 7,340 | akuzm | `medianExact` is like median but it must return a value that is present in the input set -- not sure how it's called, a medoid? The `median` function computes a real median and returns 14.
Our docs wrongly assert that these functions are synonymous, this should be fixed:
https://clickhouse.yandex/docs/en/query_lang... | 2019-10-16T10:56:04 | 2019-10-16T10:56:04 | {
"rocket": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 537,470,134 | 6,781 | filimonov | Maybe that can also be fixed by the way: https://github.com/ClickHouse/ClickHouse/issues/7174 | 2019-10-02T12:30:17 | 2019-10-02T12:30:17 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.