To all the SD card users out there

[davelacey]

All done.

https://github.com/xcore/sc_sdcard

I've tidied it up a bit but haven't tested it at all (it does compile and that's all I know - I don't have an XDK near me at the moment).

@mculibrk: If you get a github username and sign the contributors agreement (see xcore.github.com for details), I can add you as a commiter to the repo.

Regards,
Dave

[Interactive_Matter]

Cool,

thanks! Will try to get real SPI working.

BTW: What about the SD card based boot loader? Or is it too much inteleectual property right? I realy would like to use it since it makes updates on customer side much easier (esp. if you are not present at customer site ;) )

[mculibrk]

@mculibrk: If you get a github username and sign the contributors agreement (see xcore.github.com for details), I can add you as a commiter to the repo.

Done!
I'm using the same username - mculibrk

I'll start working on this shortly as I have some deadlines to catch first. :?

regards,
mculibrk

[Interactive_Matter]

I just checked the code and have little request when you are updating the code:
It nearly contains no comments in the code - so if you are analysing it - droping a line of comment or could help a lot ;)

BTW: I would also take the warning in the readme very serious: The long filename support seems to be patent protected - so we should really think if and how to integrate it to avoid patent lawsuits agains us or XMOS.

[mculibrk]

I just checked the code and have little request when you are updating the code:
It nearly contains no comments in the code - so if you are analysing it - droping a line of comment or could help a lot ;)

absolutely

BTW: I would also take the warning in the readme very serious: The long filename support seems to be patent protected - so we should really think if and how to integrate it to avoid patent lawsuits agains us or XMOS.

huh... patents, my dearest friends....
I took a quick look at the M$ FAT patent debates and as far as I can tell, the problem is in creating file name entries for long/short name combos. The Linux/Samba guys found a "shortcut" to solve the issue: here, here and here
I'll check this in more detail but if this is the "only issue" I think it will be easy enough to choose a workaround...

regards,
mculibrk

[Interactive_Matter]

…

huh... patents, my dearest friends....
I took a quick look at the M$ FAT patent debates and as far as I can tell, the problem is in creating file name entries for long/short name combos. The Linux/Samba guys found a "shortcut" to solve the issue: here, here and here
…

So there is one good thing about those patents: They force the people to be more creative ;)

I'll check this in more detail but if this is the "only issue" I think it will be easy enough to choose a workaround...

I think the rest of the problems are (hopefully) more technical - really looking forward to you contributions.

Marcus

[davelacey]

OK. So, it turns out that within XMOS towers there is some more promising (currently unreleased) sdcard + fat code that could be used as a starting point here. We'd like to put it up on github, so there are two options:

1) If you haven't done much work already you could use this as a starting point for sc_sdcard (i.e. we'd replace the current code in sc_sdcard. I think the newer version is better in practically all respects.

2) We could create a second sdcard repository for this other version.

Personally, I'd suggest 1) here since not much work has been done and there doesn't seem much need for duplication/divergence at this stage but I guess it is up to the repo maintainer. interactive-matter: Any thoughts?

Dave

[Corin]

For reference - I have written a quick port of the FatFs file system (http://elm-chan.org/fsw/ff/00index_e.html)

It's a bit slow as it just uses timers to talk to the card at the moment, but is a good a start...

Kind Regards,
Corin

[Interactive_Matter]

OK. So, it turns out that within XMOS towers there is some more promising (currently unreleased) sdcard + fat code that could be used as a starting point here. We'd like to put it up on github, so there are two options:

1) If you haven't done much work already you could use this as a starting point for sc_sdcard (i.e. we'd replace the current code in sc_sdcard. I think the newer version is better in practically all respects.

2) We could create a second sdcard repository for this other version.

Personally, I'd suggest 1) here since not much work has been done and there doesn't seem much need for duplication/divergence at this stage but I guess it is up to the repo maintainer. interactive-matter: Any thoughts?

Dave

Hi Dave,

this sounds good. Is it really completely independent of the old implementation?
The easiest would be to copy it over the old stuff and just comit - git will take care of the rest - like checking if some files are just newer versions of new files.
As long as we get a more robust library everything is a good idea ;)
I did not have time lately to look into the SD card stuff.

Are you contributor to this repo? Should I make you one?

Marcus

[Corin]

Hi Interactive_Matter,

Please make me (xmos_corin) a contributor and I'll paste it over the top.

Cheers,
Corin