You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@nuttx.apache.org by Sebastien Lorquet <se...@lorquet.fr> on 2024/10/01 08:28:03 UTC

Re: GSoC Final Report on mnemofs

Hello,

The following NAND flash chips have been ordered today:

MT29F2G01ABAGDWB-IT:G
MT29F1G01ABAFDWB-IT:F
MT29F4G01ABAFDWB-IT:F

I have selected these references because, they are biiiig and 
potentially useful, and the 2Gbit chip seems to have a different memory 
layout.

I will not be able to write new code. Specifically if it's now AI reviewed.

But I will be able to run test code of your choosing on a raspi pico or 
nucleo-l432kc

Sebastien


On 13/09/2024 22:43, Sebastien Lorquet wrote:
> Hello,
>
> Thank you for the additional information, of course I understand the 
> gsoc had to be completed in a limited time, and the current code is a 
> prototype.
>
> I think it will be cool if we manage to make it production quality for 
> all users of the nuttx codebase.
>
> I will probably send my farnell order next week so I'll have some 
> hardware to test with. the 4gbit micron chip is much larger than a nor 
> flash, and using it with a proper resilient fs will be nice.
>
> If I get help with required software, we'll make it work.
>
>
> On 9/13/24 22:09, Saurav Pal wrote:
>> Assuming the bytes are written to the flash from the start to
>> end (which I have not seen anything to suggest otherwise)
>
> <terminator voice>Theoretically.</>
>
> Joke apart, yes, it's a reasonable assumption. Sometimes it's word per 
> word not byte per byte, and even if the first bytes written could be 
> at the last address, I have never seen it done. However it is a good 
> design practice to explicitly separate the writing of the final CRC, 
> to make it obvious that you expect the previous write to be complete.
>
> The design you described should work and inspires confidence.
>
> Sebastien
>
>

Re: GSoC Final Report on mnemofs

Posted by "Alan C. Assis" <ac...@gmail.com>.
AFAIK the AI bot is used only to help identify things that are missing in
the PR, so it avoids asking users again and again to fill the Summary, etc.

BTW, I don't think simple things like that need Apache or PMC approval.

It is an open-source project after all and contributions are welcome!

But I'm glad we have you to disagree, it means we are doing the right thing
("Unanimity is always stupid." -- Nelson Rodrigues).

BR,

Alan

On Thu, Oct 3, 2024 at 5:50 AM Sebastien Lorquet <se...@lorquet.fr>
wrote:

> Hi,
>
> I am now the owner of 4 Micron NAND chips and I am at the disposal of
> NuttX developers who want to test code to drive and use them.
>
>
> Tomek: It doesnt change a damn thing.
>
> If LLM are used to make up for contributor laziness that's even worse.
> You'll get unreliable slop when the proper thing to do is close these
> bogus requests and ask contributors to fill actual required information.
>
> Was the use of ai discussed and voted by the project? No need, I'll be
> the only one against it?
>
> Sebastien
>
>
> On 01/10/2024 13:01, Tomek CEDRO wrote:
> > On Tue, Oct 1, 2024 at 10:28 AM Sebastien Lorquet <se...@lorquet.fr>
> wrote:
> >> (..)
> >> I will not be able to write new code. Specifically if it's now AI
> reviewed.
> > Sebastien the code is not reviewed by the AI it only provides initial
> > feedback on the PR and commit descriptions something we are constantly
> > repeating and asking some reporters to fill in with meaningful content
> > and what should be done correctly in the first place but its often not
> > :-) Humans are still reviewing, commenting, requesting changes,
> > approving, and merging the code no worries :-)
> >
>

Re: GSoC Final Report on mnemofs

Posted by Sebastien Lorquet <se...@lorquet.fr>.
Hi,

I am now the owner of 4 Micron NAND chips and I am at the disposal of 
NuttX developers who want to test code to drive and use them.


Tomek: It doesnt change a damn thing.

If LLM are used to make up for contributor laziness that's even worse. 
You'll get unreliable slop when the proper thing to do is close these 
bogus requests and ask contributors to fill actual required information.

Was the use of ai discussed and voted by the project? No need, I'll be 
the only one against it?

Sebastien


On 01/10/2024 13:01, Tomek CEDRO wrote:
> On Tue, Oct 1, 2024 at 10:28 AM Sebastien Lorquet <se...@lorquet.fr> wrote:
>> (..)
>> I will not be able to write new code. Specifically if it's now AI reviewed.
> Sebastien the code is not reviewed by the AI it only provides initial
> feedback on the PR and commit descriptions something we are constantly
> repeating and asking some reporters to fill in with meaningful content
> and what should be done correctly in the first place but its often not
> :-) Humans are still reviewing, commenting, requesting changes,
> approving, and merging the code no worries :-)
>

Re: GSoC Final Report on mnemofs

Posted by Tomek CEDRO <to...@cedro.info>.
On Tue, Oct 1, 2024 at 10:28 AM Sebastien Lorquet <se...@lorquet.fr> wrote:
> (..)
> I will not be able to write new code. Specifically if it's now AI reviewed.

Sebastien the code is not reviewed by the AI it only provides initial
feedback on the PR and commit descriptions something we are constantly
repeating and asking some reporters to fill in with meaningful content
and what should be done correctly in the first place but its often not
:-) Humans are still reviewing, commenting, requesting changes,
approving, and merging the code no worries :-)

-- 
CeDeROM, SQ7MHZ, http://www.tomek.cedro.info