Sand Fox (Posts about Misc)

Generating UUIDv7 in PostgreSQL 18

2025-10-02T00:37:00+03:00

Another post just turned from a useful algorithm to madskillz. Since PostgreSQL 18 the method is just:

select uuidv7();

A Case for JSON6

2025-04-19T06:43:00+03:00

Just a required follow up for the previous post: JSON5 sucks too, like any other possible format. Just look at this:

{
    certificate: "\
-----BEGIN PRIVATE KEY-----\n\
MIIBVQIBADANBgkqhkiG9w0BAQEFAASCAT8wggE7AgEAAkEAt6wGEsfdarMo9KET\n\
sNv9gpBWkGfhIq3/Jhr14+3d8TXprmTk/XWReo5DA3+SbB7ebF3xUtN/4K5lUkcG\n\
1AbdywIDAQABAkBuMbfnFqAhvuFoeydMKYAcECrPMnOhEdENdIlnuTM53kBf18f/\n\
KaWRWv8ViuZ31GpArY8cBzj0YG30vE49Il8ZAiEA7PpotzcpjNcMKnwEBm3q0c6H\n\
3RRZD6i32Q/vXcQBeB0CIQDGaj4BJ/fuqiHvKzetfjhpOD+sYr3VDgEyjTcojhzZ\n\
BwIgTBRYafmbrUuc7EbERAwlxxW3KJLPxOc1nsou3rt+fUECIQCjBGUfJAyDDUtG\n\
KpbbN0n3wRwncRUQuCnps7Zu3pv6/wIhAIIcKty7O0T7auctlpMSdyw6vp0w818x\n\
namFjg3cbBs4\n\
-----END PRIVATE KEY-----\n"
}

Read more… (1 min remaining to read)

Is JSON5 a better JSON?

2025-04-16T06:10:00+03:00

Yes it is, at least when it comes to configuration files.

As I said earlier, JSON is a good serialization language but a terrible config language. JSON5 builds on its strong sides: it's backwards compatible with JSON and is still a subset of JavaScript and it is still relatively easy to parse. Then it adds all the good stuff like unquoted keys, trailing commas, and a must-have thing for any config, comments. For me it has only two real downsides, both really minor. First is, as most config-oriented languages, it's tricky for automated modification, a feature it shares with TOML and YAML. The second one is JSON6, while ES6 syntax sugar is nice, it's just a 15th competing standard, and the worst, competing with itself.

Ignoring SSL Errors in Chrome

2024-05-07T17:45:00+03:00

Very useful when you are developing anything WebAuthn related.

google-chrome --ignore-certificate-errors --user-data-dir=$HOME/tmp/chrome-dev

--ignore-certificate-errors will disable any SSL warnings like self-signed certs.

--user-data-dir will isolate this unsafe Chrome instance from your regular browsing.

Generating UUIDv7 in PostgreSQL

2025-10-02T00:43:00+03:00

But what about Postgres? We can do it too. Hopefully generators for new UUID versions will be added to uuid-ossp but it is not yet the case.

The easiest way for now is to use pgcrypto, so let's start with that:

create extension pgcrypto;

Now port the solution from the MySQL post:

select
    lpad(to_hex((extract(epoch from now()) * 1000)::bigint), '12', '0') ||
    '7' || -- version
    substr(encode(gen_random_bytes(2), 'hex'), 2) ||
    to_hex((floor(random() * 4) + 8)::int) || -- variant bits
    substr(encode(gen_random_bytes(8), 'hex'), 2)
    as uuid;

Like with MariaDB, Postgres doesn't care about dashes on insertion, so this value can be directly applied to a field with the uuid type.

And again, to make it a formatted value, use a type cast:

select
    (lpad(to_hex((extract(epoch from now()) * 1000)::bigint), '12', '0') ||
    '7' ||
    substr(encode(gen_random_bytes(2), 'hex'), 2) ||
    to_hex((floor(random() * 4) + 8)::int) ||
    substr(encode(gen_random_bytes(8), 'hex'), 2))::uuid
    as uuid;

And to have a bytea value, use decode:

select
    decode(
        lpad(to_hex((extract(epoch from now()) * 1000)::bigint), '12', '0') ||
        '7' ||
        substr(encode(gen_random_bytes(2), 'hex'), 2) ||
        to_hex((floor(random() * 4) + 8)::int) ||
        substr(encode(gen_random_bytes(8), 'hex'), 2),
        'hex'
    ) as uuid;

Generating UUIDv7 in MySQL and MariaDB

2024-01-20T07:06:00+02:00

A quick way to generate UUIDv7 in MySQL 8 and MariaDB 11:

select
    concat(
        -- Use NOW(3) to get milliseconds
        lpad(hex(unix_timestamp(now(3)) * 1000), 12, '0'),
        '7', -- version
        substr(hex(random_bytes(2)), 2),
        hex(floor(rand() * 4 + 8)), -- variant bits
        substr(hex(random_bytes(8)), 2)
   ) as uuid;

This will generate a hex string that can be inserted into MariaDB's UUID type as is, dashes will be added automatically.

If you need it as a binary value, wrap with unhex:

select
    unhex(concat(
        lpad(hex(unix_timestamp(now(3)) * 1000), 12, '0'),
        '7',
        substr(hex(random_bytes(2)), 2),
        hex(floor(rand() * 4 + 8)),
        substr(hex(random_bytes(8)), 2)
    )) as uuid;

If you specifically need a valid RFC 4122 string representation, use a bin formatter function (MySQL) or a type cast (MariaDB):

-- MySQL
select
    bin_to_uuid(unhex(concat(
        lpad(hex(unix_timestamp(now(3)) * 1000), 12, '0'),
        '7',
        substr(hex(random_bytes(2)), 2),
        hex(floor(rand() * 4 + 8)),
        substr(hex(random_bytes(8)), 2)
    ))) as uuid;

-- MariaDB
select
    cast((concat(
        lpad(hex(unix_timestamp(now(3)) * 1000), 12, '0'),
        '7',
        substr(hex(random_bytes(2)), 2),
        hex(floor(rand() * 4 + 8)),
        substr(hex(random_bytes(8)), 2)
    )) as uuid) as uuid;

UPD: Postgres can do it too!

Running 32-bit Images in GitLab CI

2023-09-07T16:03:00+03:00

A thing that was not obvious to me and relatively hard to google was a way to run a 32-bit docker image in the GitLab CI. Docker itself has --platform param but it is not exposed to the .gitlab-ci.yml file.

Luckily it's easy for the official Docker registry. Apart from auto-selection images, it also contains per-architecture image repos with distinct package names. You only need to specify a desired architecture prefix: like amd64/php or i386/php.

Example:

.test:
  script:
    - composer update
    - vendor/bin/phpunit

test-64bit:
  extends: .test
  stage: test
  image: amd64/php:8

test-32bit:
  extends: .test
  stage: test
  image: i386/php:8

or using the matrix:

test:
  stage: test
  image: ${ARCH}/php:${PHP_VERSION}
  script:
    - composer update
    - vendor/bin/phpunit
  parallel:
    matrix:
      - PHP_VERSION:
          - '7'
          - '8'
        ARCH:
          - amd64
          - i386

Resizing small EFI partition

2023-04-17T15:46:00+03:00

It started with the firmware failing to update because the EFI partition was low on free space. (Dell, why do you need 70 MB on the EFI while also making it as small as 150 MB at the same time?) Well, resizing the partition is the obvious solution except it ended in a wonderful error message:

GNU parted cannot resize this partition to this size. We're working on it!

(spoiler: they're not)

In my case the GPT partition was resized, just the filesystem was not grown, I only needed to reformat it.

Read more… (1 min remaining to read)

GPL Incompatibility

2023-02-22T16:28:00+02:00

"GPL incompatibility" of a FOSS license is not a problem of the incompatible license, it's the problem of the GPL. Most permissive licenses and file-based copyleft licenses can be used together in a project without any conflict and that's great. It's GPL that doesn't like any neighbors.

LGPL is a bad practice

2024-11-24T21:26:00+02:00

TL;DR: Don't use LGPL, avoid library-based copyleft in general.

The closest acceptable no-brainer replacement should be GNU GPL with Classpath Exception, but the concept of library-based copyleft itself is flawed. It's better to consider MPL (less copyleft) or full GPL (more copyleft).

Read more… (1 min remaining to read)

Python Type Hints

2022-06-22T04:52:00+03:00

Python type hints are garbage.

No Discrimination

2022-03-16T19:45:00+02:00

In these uneasy times I just want to remind to all these passionate activists around that there are other ethical points beyond any given "today thing" even when the today thing is truly horrible.

The Open Source Definition states that open-source software must comply with the following criteria:

5. No Discrimination Against Persons or Groups The license must not discriminate against any person or group of persons.

6. No Discrimination Against Fields of Endeavor The license must not restrict anyone from making use of the program in a specific field of endeavor. For example, it may not restrict the program from being used in a business, or from being used for genetic research.

So whether you are breaking these rules for fun or showing your political position, you should remember:

It is not as fun as it seems.
You are not hurting the people you hate, you are hurting the open-source community.

sass-hsv

2021-08-28T05:56:00+03:00

I haven't found a nice small helper for HSV/HSB color format like the one LESS has so I published my first npm package.

https://sandfox.dev/js/sass-hsv.html

TOML vs YAML

2021-08-10T03:41:00+03:00

So... Let's compare.

Open Letter in Support of Richard Stallman

2021-03-24T15:51:00+02:00

https://rms-support-letter.github.io

Please even if you do not agree with him politically but still don't like that raging mobs decide for others, sign the letter here: https://github.com/rms-support-letter/rms-support-letter.github.io/pulls

OTP Links

2020-08-26T16:43:00+03:00

Have you ever seen an one time password link in a regular <a> tag?

Here is an example TOTP link from Google Authenticator repo:

otpauth://totp/Example:alice@google.com?secret=JBSWY3DPEHPK3PXP&issuer=Example

The most familiar form for them is QR codes:

Which is a problem if you're alterady on your mobile device and you can't scan it. For this case there is usually a text like "Please manually add the key JBSW Y3DP EHPK 3PXP to your Authenticator app". But what about the link itself? Well, just try it on your phone: click me.

I'm really shocked that it works perfectly and that I have never seen this simple and very web style solution in action.

Is NEON a Better YAML?

2020-06-15T22:18:00+03:00

Short answer: NO. If you know what I mean.

Read more… (1 min remaining to read)

What's Wrong with StrictYAML

2020-05-29T05:48:00+03:00

While I'm still delaying my position on TOML vs YAML holy war let's look at the StrictYAML.

StrictYAML is a type-safe YAML parser that parses and validates a restricted subset of the YAML specification.

Seems great but let's look at the removed features list, Implicit Typing to be precise:

Read more… (1 min remaining to read)

Free Domain Names

2020-04-07T09:23:00+03:00

Just created a list of free domain name zones that still work if you need one for whatever reason.

You can find it here: freedomainzones.netlify.app.

I Don't Like JavaScript

2020-04-01T00:34:00+03:00

When you created a cool tool in a language you don't like.

Read more… (1 min remaining to read)

Small Personal Update

2020-03-28T01:16:00+02:00

A small personal update about services and social networks.

I changed my username on GitHub and Packagist to @arokettu. I'm also on Keybase now.

GitHub Signatures in Gitea

2019-11-15T02:16:00+02:00

A small lifehack for the Gitea owners. As you may know, GitHub signs commits done with its web interface with its GPG key. These commits may end up on your Gitea server. It looks rather unconvincing:

Read more… (1 min remaining to read)

Dark Theme for the Blog

2019-10-26T15:40:00+03:00

Just finalized the dark theme for the blog. The idea is simple, use prefers-color-scheme media query and wrap dark style elements with it:

@media (prefers-color-scheme: dark) {
    a {
        color: #4be
    }
}

Read more… (1 min remaining to read)

On Common Misuse of the File Formats

2019-10-25T18:20:00+03:00

A small rant about text file format holy wars. I believe that most of the file formats that participate in these holy wars are rather good and the most of the frustration comes from their misuse.

Read more… (1 min remaining to read)

Bitbucket to Remove Mercurial Support

2019-08-20T21:40:00+03:00

So Mercurial lost its biggest hosting platform. Sad to see it because I always thought that it has much more usable CLI and much better branching system for corporate development than git.

After much consideration, we've decided to remove Mercurial support from Bitbucket Cloud and its API. Mercurial features and repositories will be officially removed from Bitbucket and its API on June 1, 2020.

https://bitbucket.org/blog/sunsetting-mercurial-support-in-bitbucket

sandfox.dev

2019-04-19T17:35:00+03:00

I created a website to host documentation for my public libraries and to serve as a primary index for them

https://sandfox.dev/

Created with Sphinx with a small script that collects docs directories from project repos

Rejecting Contributor Covenant

2019-02-17T22:44:00+02:00

Partially as a follow-up for a previous post. As we now know, the sole purpose of the Contributor Covenant is silencing SJW's political enemies. Luckily Coraline admitted that CC is political so resisting it is also mere politics :D

Read more… (1 min remaining to read)

Political Censorship on GitHub

2019-02-17T18:58:00+02:00

I never expected this blog to go political but the situation became outrageous

Read more… (1 min remaining to read)

Farewell to Jekyll

2020-05-28T07:41:00+03:00

Just ported the blog to Pelican and removed all Jekyll files. Currently it seems the most logical choice for me

First, I started to use reST much more extensively than Markdown and I want a blog generator that supports it better

Second is my new interest in Python

Third, Pelican seems to have much more features that I like compared to Jekyll and other static generators that I tested, especially Python ones (however Nikola also seems a good alternative but I had to stop at something)

UPD (May 2020): Nikola feels waaay better.

Harold Finch Uses JetBrains IDE

2016-06-16T03:20:00+03:00

Yay, Harold Finch from the Person of Interest was spotted using some JetBrains IDE.

Read more… (1 min remaining to read)

RubyMine Supports Markdown

2016-04-06T02:55:00+03:00

Wow. It seems RubyMine and other JetBrains products now have very nice Markdown support. I haven't even noticed it until last article. (I usually had MultiMarkdown plugin installed)

Killer feature for me is that it allows working with code blocks of known languages as if it was real code, even have code completion and some basic inspections there.

Kinda Blog

2016-03-09T17:08:49+03:00

In the beginning there was a Gemfile. And the Gemfile contained gem 'jekyll', '~> 3.1'

I may post some earlier notes with earlier dates