Sd File::availableForWrite()

[fabianoriccardi]

Hi, I have recently discovered int availableForWrite() to check the amount of available bytes without waiting for new "sector" (hence avoiding blocking writings). here) the example that demonstrate it.

Is there any plan to add to esp8266 core?

[earlephilhower]

@fabiuz7, interesting note. I'm not sure that the same technique can be used by our newer version of the SdFat library (the Arduino fork is ancient...), but if it is then it seems like it would be a reasonably simple PR. Would you be interested in submitting one?

[JAndrassy]

start by adding availableForWrite to Print
https://github.com/arduino/ArduinoCore-API/blob/76e931e97e378a4a4ba3ea12dd8a09a1b16d1286/api/Print.h#L59

[fabianoriccardi]

I may try, but I'm not very familiar with sdfat library. However I was looking at esp8266 arduino repo, and I found that Serial and WiFi already as a similar method. Does a virtual "availableForWrite" in Print class introduce conflict with WiFi and Serial classes?

[earlephilhower]

Good pointer, @JAndrassy! Looks like that was added ~1yr ago to the Arduino upstream core.

@fabiuz7 In the end if there's no compile-time ambiguity, it should be fine. I think that's what would be needed, anyway. You can add it in a PR and the continuous integration system will verify no compiles get broken.

[JAndrassy]

Good pointer, @JAndrassy! Looks like that was added ~1yr ago to the Arduino upstream core.

I linked to API repo which is not really used. it missed it too. to AVR and SAMD it was added in 2017

[earlephilhower]

@fabiuz7 can you give #7658 a try and report back? I don't have my SD card reader hooked up to verify the new call.

[fabianoriccardi]

Hi, I had tried this sketch. It compiles but it continues to store the data in the local buffer, the SD card is never written. availableForWrite return always 0. Is it intended?

[earlephilhower]

Hi, I had tried this sketch. It compiles but it continues to store the data in the local buffer, the SD card is never written. availableForWrite return always 0. Is it intended?

Thanks for the feedback. I double checked (added the avail4Write call to LittleFS and got expected results) the code and it's actually working correctly.

The latest SDFat library does not support availableForWrite (see https://github.com/greiman/SdFat ), so that sketch can't work on it. It looks like the Arduino guys preserved the 7(!) year old fork they took and have added the call to it. So, on the 8266 it's calling the default "return 0" handler.

I don't know why Arduino would have added new stuff to the ancient fork vs. pulling in the currently maintained version and doing a PR.

https://github.com/arduino-libraries/SD/blame/85dbcca432d1b658d0d848f5f7ba0a9e811afc04/src/utility/SdFile.cpp#L1492

Let me see if the hack they did can be added to our own ESP8266SdFat library. I'm not so sure, given their ancient codebase, but will let you know if something comes up.

[earlephilhower]

I've spent some time trying to hack in the Arduino patch to the modern SD library and have to say that it's beaten me, @fabiuz7 . :(

There have been a lot of little changes to the SD library architecture which make some of the hacks Arduino did not directly translatable. So, I think you's out of luck for SD.availableForWrite() for the time being. The actual patch to consolidate it under a Print class, however, should still go through if only for consistency.

[fabianoriccardi]

Thanks for the investigation. That's a pity that Arduino's version diverged so much from the original SdFat, limiting the inclusion of features in other cores :(

[fabianoriccardi]

Hi again,
I digged a bit into sdfat library, and I discovered the old isBusy() method. Is it the core method to implement availableForWrite, isn't it? This method is pretty old, and it can be found in the fork of @earlephilhower, so maybe there is a change to implement it, or at least to avoid long sporadic writings.

[earlephilhower]

See #7779