Just sort the keys first!

Example:

cat a.json | jq --sort-keys . > aa.json
cat b.json | jq --sort-keys . > bb.json
vimdiff aa.json bb.json

Server side pagination is intrinsically not accurate, as long as the data is dynamic.

The data items could be inserted, deleted or changed on the server side while the user goes forward and backward among the pages.

However, there is an algorithm that can keep the pagination as stable as possible:

1. encode the id and sorting fields of last value in a page as the continue-token
2. return the continue-token along with each page
3. the client must pass the continue-token to fetch the next page
4. the next page starts with value > continue-token || (value == continue-token && value.id > continue-token.id)

Reference

• Pagination with Dynamic Data Loading

A BUID is a 128-bit unique ID composed of two 64-bit parts: shard and key.

It is not only a unique ID, but also contains the sharding information, so that the messages with the same BUID could be stored together within the same DB shard.

Also, when a message is stored in a shard, the shard part of the BUID can be trimmed off to save the space, and only the key part needs to be stored as the primary key.

Bigendian is chosen to make each part byte-wise lexicographic sortable.

The string representation uses basex 62 encoding.

curl [args] --write-out "%{time_total}s" --output /dev/null --silent [URL] 

git fetch       --prune  origin
git branch      --delete [branch-name]
git push origin --delete [branch-name]

wget --recursive --level=2 --no-parent --no-clobber --convert-links --continue
[URL]

sudo apt install byobu
sudo purge-old-kernels

Reference:

• Remove Old Kernels (https://help.ubuntu.com/community/Lubuntu/Documentation/RemoveOldKernels)