Byteserver project status

Éloi Rivard

unread,

Apr 17, 2019, 11:21:24 AM4/17/19

to zodb

Hello,

I was wondering about the byteserver project status. The performance graph looks very promising, but the project has not received attention since 3 years (there is also this pull request waiting for a few months).

Have you plans about developping this project, or was it just a proof of concept?

Did you encounter technical difficulties?

Would you accept help to make it progress? If so, maybe creating exhaustive issues for the missing features would be a good point to start? What do you think?

Best regards,

Éloi

Éloi Rivard

unread,

Apr 17, 2019, 4:12:45 PM4/17/19

to zodb

To give you a bit more context: my company has regulary performance issues. As we have some Rust developpers, we were wondering if contributing to this project was a good option for us.

Jim Fulton

unread,

Jun 23, 2019, 4:03:05 PM6/23/19

to Éloi Rivard, zodb

On Wed, Apr 17, 2019 at 9:21 AM Éloi Rivard <azm...@gmail.com> wrote:

Hello,

Apologies for the very late reply.

I was wondering about the byteserver project status. The performance graph looks very promising,

It was a proof of concept that worked out well.

but the project has not received attention since 3 years (there is also this pull request waiting for a few months).

Oops. :(

Have you plans about developping this project, or was it just a proof of concept?

Well, I thought I would take it further. I even added support for using msgpack for the ZEO protocol and client-side conflict resolution to ZEO to enable it.

Consulting work took me in a different direction.

Did you encounter technical difficulties?

No, at least nothing that more digging wouldn't address.

Would you accept help to make it progress? If so, maybe creating exhaustive issues for the missing features would be a good point to start? What do you think?

I had plans for a new file-storage that addressed a number of issues in the current file storage implemented as part of this work. See https://github.com/jimfulton/filestorage2. Some major benefits:

Use less memory by leveraging memory mapping.
Make packing more incremental and less disruptive.

Since then, I've had opportunity to get more experience with memory mapping which makes the filestorage2 ideas even more appealing to me.

Unfortunately, it's been a while since I've been able to use ZODB in my day job :(, which makes it difficult for me to work on it. I'm motivated and informed by solving actual problems.