clickhouse / clickhouse-docs Goto Github PK

The hasToken function is not documented: https://github.com/ClickHouse/ClickHouse/pull/6596/files

Some of the files are blocked:

Based on comments here: https://clickhouse-inc.slack.com/archives/C03D1A18GCB/p1652203883224569

We have a MySQL page: https://clickhouse.com/docs/en/interfaces/mysql/, but nothing on Postgres. Plus a user guide demonstrating both would be helpful.

The JDBC url requires a special port and ssl parameter in order to connect to ClickHouse cloud.
And in slack channel for private preview customer the customer specifically asked for help because it is not so obvious how to connect to our cloud.

Therefore let's add a special cloud section to the Connecting Applications to ClickHouse with JDBC doc.

• use screenshots from the control plane with the https connection values
• give concrete java example program, e.g.

import com.clickhouse.jdbc.ClickHouseDataSource;

import java.sql.Connection;
import java.sql.ResultSet;
import java.sql.ResultSetMetaData;
import java.sql.Statement;

// connect to cloud instance
public class JDBC2CHCloud {
    public static void main(String[] args) throws Exception {

        String url = "jdbc:ch://yxaeq2tmh3.eu-west-1.aws.clickhouse.cloud:8443/default?ssl=true";

        ClickHouseDataSource dataSource = new ClickHouseDataSource(url);
        try (Connection connection = dataSource.getConnection("username", "pw");
             Statement statement = connection.createStatement();
             ResultSet resultSet = statement.executeQuery(" select * from system.tables limit 10")) {
            ResultSetMetaData resultSetMetaData = resultSet.getMetaData();
            int columns = resultSetMetaData.getColumnCount();
            while (resultSet.next()) {
                for (int c = 1; c <= columns; c++) {
                    System.out.print(resultSetMetaData.getColumnName(c) + ":" + resultSet.getString(c) + (c < columns ? ", " : "\n"));
                }
            }
        }
    }
}

The changelog.md file is auto-generated and is not in the correct location for Docusaurus.

Also, the script generates a TOC that is no longer needed.

The svg function is not documented. PR is here: https://github.com/ClickHouse/ClickHouse/pull/11871/files#diff-d7b3a724f1529fd06c9bd22155f848d82cad53b24249b180c00b2590466f7adf

• How to choose a primary key?
• How to add a column in ClickHouse?
• Too many parts
• How to filter ClickHouse table by an array column contents?
• How to insert all rows from one table to another of identical structure?
• How to kill a process (query) in ClickHouse?
• How to implement pivot (like in pandas)?
• How to remove the default ClickHouse user through users.d?
• Importing MySQL dump to ClickHouse
• Window function workarounds (row_number, lag/lead, running diff/sum/average)

https://stackoverflow.com/questions/72197533/how-to-set-the-secret-tag-in-the-remote-servers-tag

I am not a GitHub expert, but this looks to me like the workflow failed because my PR was based on a fork and not a local branch:

++ git branch --show-current
+ GIT_BRANCH=main
+ '[' reorder-example-datasets ']'
+ '[' reorder-example-datasets == main ']'
+ git fetch origin --depth=1 -- reorder-example-datasets:reorder-example-datasets
fatal: couldn't find remote ref reorder-example-datasets

I replaced the PR with another based on a branch created in this repo and it worked fine.

Named collections work for Kafka, but there is not an example in the docs. There is a test in the PR though: https://github.com/ClickHouse/ClickHouse/blob/b39c19399ff951a598dd35e575c715c5c4dfb5a2/tests/integration/test_storage_kafka/configs/named_collection.xml

The page about CMake flags for building ClickHouse appears to be missing.

Specifically:

• DevDocs https://devdocs.io/
• Dash https://kapeli.com/dash

Keep ZooKeeper as obsolete option.

We used to have a separate repo with images for blog (immediately served via GitHub Pages on a separate domain without any CI/CD), we could do the same with guides images

The existing 404.html page appears to be using assets from the old /docs site (that is my theory at least)

Experimental Feature, transaction support:
https://clickhouse.com/docs/en/whats-new/changelog/#:~:text=Added%20support%20for%20transactions%20for%20simple%20MergeTree%20tables.%20This%20feature%20is%20highly%20experimental%20and%20not%20recommended%20for%20production.%20Part%20of%20%2322086.%20%2324258%20(tavplubix).

The screenshots on https://clickhouse.com/docs/en/integrations/dbt/dbt-connecting/ could be of higher quality, and the images are never scaled with the page. It is hard to look at them on wide screen

How to fix ZooKeeper issues where we need to detach and reload the parts. And also the issue where CH won't start because of an issue with a table definition in the /metadata folder

These three files were updated in the wrong format:

docs/en/sql-reference/functions/date-time-functions.md | 7 ++++
docs/en/sql-reference/functions/other-functions.md | 38 ++++++++++++++++++++
docs/ru/sql-reference/table-functions/postgresql.md | 2 +-

It is currently /en/, but we could consider changing it to /references. It would require a very large number of redirect files.

Feedback from some experienced users - complain about markdown of https://clickhouse.com/docs/en/development. About 10% of space used, a lot of half-empty blocks, huge titles font and tiny text font.

There is room on a wide screen, but probably needs some custom CSS:

Images on the Kafka Integration guide look blurry.

Similar to #83

Once example:

You can open a menu item by clicking either on text or on the arrow. But to close menu item it's only possible to click on arrow.

The docs are served up from two different repos, so the "edit" button is non-trivial. We will probably have to swizzle the component and customize.

Looks like Docusaurus has a plugin for it: https://www.npmjs.com/package/docusaurus-plugin-segment

In Twitter, we received the following enhancement request:

suggestion: indicate in the docs in which versions things became available. For example: doesn't say when date_diff became available. groupArraySorted also showed up without a version indication even though it was only in master (was since reverted entirely)

https://twitter.com/roncohen/status/1526572164723392512

This URL is not displaying correctly for me:
https://clickhouse.com/docs/quick-start/

I think it should probably redirect to:
https://clickhouse.com/docs/en/quick-start/

However, this URL is still found in Google Search for "clickhouse quick start":

Result upon clicking on it - land on an odd-looking page

The table listing all our supported in and output formats is broken since moving to Docusaurus.

@qoega I did not remove the partition by as we mention using the partitions when parallelizing writes to s3:

Note the use of partitioning on the pickup_date field. Whilst usually a technique to assist with data management, we can later use this key to parallelize writes to s3.

Can you please pop a note in here and we can discuss whether or not to make a change?

https://clickhouse.com/docs/en/getting-started/example-datasets/metrica#obtaining-tables-from-prepared-partitions

Here it's unclear what does it mean "fix permissions if necessary" and looks like it won't work with Atomic database (which is by default long time ago).

Add all the remaining entries from https://news.ycombinator.com/item?id=30878761
and also all the ones discussed in Slack.

I mean, they're ok, but they are very Docusarus-y...let's try and style them to something more appropriate for ClickHouse.

PRs are ClickHouse/ClickHouse@731d768 and ClickHouse/ClickHouse#3139

And also we should reorder to put the most interesting datasets first. (Let's put UK Property Price Paid first)

We should consider modifying the CSS so that we can enable dark mode but still keep the CH look-and-feel. Who knows, it might look OK out-of-the-box! But I suspect we will need some design effort.

selecting documents in header at base clickhouse.com directs to https://clickhouse.com/docs/en/ instead of new docs page.
Not sure if this is intentional by design or if the link is supposed to go to new docs landing page https://clickhouse.com/docs/intro, etc.

Some settings can be applied to specific queries using the SETTINGS keywords or the the current session using SET (eg. max_insert_threads). On the other hand, we can have settings that that must go into the profile and only takes effect on server restarts (eg. http_max_field_value_size , #36783).

We should distinguish these types of settings in the docs either grouping them into categories, or labelling the ones that require a server restart

based on release 22.2.X
https://clickhouse.com/docs/en/whats-new/changelog/#:~:text=Add%20a%20setting,Igor%20Nikonov

Bottom end of the left menu. It does not combine well with the footer that is very light blue. The low contrast of white and very light blue does not combine well with the rounded corner of dark blue menu.

It appears to have stopped working about a week ago (so even for the old docs)

Specifically, an issue was found: search by “jdbc” can’t find https://clickhouse.com/docs/integrations/jdbc

The highlighted color could have more contrast:

One of our students asked a question in the Bootcamp about the case sensitivity of the ORDER BY clause on String values, related to this page:
https://clickhouse.com/docs/en/sql-reference/statements/select/order-by/

The assumption is that the ASCII order for Strings is used, although not specifically stated on this page. It is noted on this page that when using COLLATE, sorting is case-insensitive:
When using COLLATE, sorting is always case-insensitive.

However, to make this specifically clear, suggesting that we add a statement on the default String sort order.

In looking at the way this is laid out, I'm wondering if we might better prepare for future growth by making the default location, in English, to the following:

https://clickhouse.com/docs/en/

In doing this, we could then essentially do a new build of the docs for each existing language:

https://clickhouse.com/docs/ru/
https://clickhouse.com/docs/zh/
https://clickhouse.com/docs/ja/

And in the near future, if we were to use DeepL or others to use machine translation:

https://clickhouse.com/docs/fr/
https://clickhouse.com/docs/kr/
https://clickhouse.com/docs/es/
https://clickhouse.com/docs/de/
...

This might mean building the whole Docusaurus set for each language, I'm not sure how that works exactly. Ideally from the default language page (redirected from https://clickhouse.com/docs/), there would be a language selection option in the upper right corner.

address applying aggregation on insertion of data

They have a built-in connector for ClickHouse, so it's fairly straightforward to get started: https://arctype.com/

It uses chmod 666, but should be chmod 600