Update WebServer.h - #49

As my recent commits have caused a couple of your issues (my apologies), here are my comments from a quick skim of the code:

• The signature of write added back in this commit doesn't implement buffering correctly.
• The same signature of write added back in this commit has been added to the Print class in a recent commit to Arduino 1.5. It would be best to use the same code as this for consistency (ideally conditionally compiled against the version number so that our code doesn't duplicate the Print version when available).

Am away from my PC at the moment so I can't dig out the relevant commit but can do this evening if that would be useful?

Now I'm back at the PC I don't have to rely on my flaky memory...

The write(const uint8_t *buffer, size_t size) method has been implemented in the base Print class since at least 2011 according to https://github.com/arduino/Arduino/blob/d2a38e4b5a4260a70b79276667d7e64ce6547010/hardware/arduino/cores/arduino/Print.cpp so I would have thought that we shouldn't need to implement it ourselves - am I getting the wrong end of the stick? (The recent commit I was referring to above concerned signed chars not unsigned so please ignore that point).

I saw that Print had the method implemented but it was not compiling for me
so I went ahead an implemented it myself since anyways the default
implementation copies every character and I wanted to push entire buffers
(I'm doing file transfers).

Now that you asked about why is not picking up the base implementation I
found this:

http://forum.arduino.cc/index.php/topic,107400.0.html

Basically if you want the compiler to look for overloaded virtual functions
you need to add "using Print::write;" (it works for me). Interesting....I
think I have never run into this case.

Matias

On Mon, Feb 3, 2014 at 5:37 PM, Matt Robinson [email protected]:

Now I'm back at the PC I don't have to rely on my flaky memory...

The write(const uint8_t *buffer, size_t size) method has been implemented
in the base Print class since at least 2011 according to
https://github.com/arduino/Arduino/blob/d2a38e4b5a4260a70b79276667d7e64ce6547010/hardware/arduino/cores/arduino/Print.cppso I would have thought that we shouldn't need to implement it ourselves -
am I getting the wrong end of the stick? (The recent commit I was referring
to above concerned signed chars not unsigned so please ignore that point).

Reply to this email directly or view it on GitHubhttps://github.com//pull/49#issuecomment-33997054
.

Well, you learn something every day... Is there a noticeable improvement in the file transfer speed using your implementation vs the default?

I did not run any performance test on it yet. My approach was more pragmatical. I saw that the ethernet client had support for block transfers so I used it. I think that on my particular set up it won't matter too much since the network card is using SPI anyways (it will just save some instructions).

Thanks

Just an idea: instead of Print the server could implement the Stream interface and then you could directly parse the request body with the aJson library.

Thanks for all the hard work you guys have put into this library.