>> This inspired me to propose dropping 32bit support for PostgreSQL starting with PG17.
>I don't think this is a great idea. Even if Intel isn't interested,
>there'll be plenty of 32-bit left in the lower end of the market
>(think ARM, IoT, and so on).
Hello Tom,
Certainly there are many 32bit implementations out there in many markets.
Therefore I spoke from inspiration by Intel's move, what does not implicitely indicate a preference of x86 architecture.
Considering ARM (worlds most used ISA) I think we should focus on the use cases.
All implementations in many various scenarios (keyboard controllers,SetTop-Boxes,TV sets, SSD/Hard-disk-controllers,BMC controllers and many more) are no real use cases for Postgres on ARM.
The only relevant usage scenario at this time is in the datacenter when AARCH64 based CPU designs are used in servers.
The most spreaded usage of ARM (nowadays almost 64bit), which is used dayly by biliion of people, are the Smartphones and tablets, which are completelely unsupported by Postgres!
An officially supported access module for remote access to server-based database would bring Postgres to a much broader knowledge and use for many people.
(why not provide an official libpq on IOS or Android to enable specialized client applications?)
On the IoT side I have not much knowledge, perhaps you have a relevant example of a native 32bit server implementation in this area. But to my knowledge there is no special support in our code base yet (OS, File systems, storage, administration).
For me the central question is:
Are there many use cases for new PG server installations on 32bit starting with late 2024/2025?
and not
Do we provide support for NEW VERSIONS for aging architectures which themselves aren't used for new installations any more and mostly never are updated even to currently supported versions?
Hans Buschmann
