I wasn't able to enable mail delivery in my account, so I'm reposting
my response here.

---
From: Daniel Stenberg <daniel_at_haxx.se>
Date: Fri, 28 Aug 2009 23:28:59 +0200 (CEST)
>> The nature of the program I'm making is that I need to be able to get the
>> content-length header to allocate space for an image, then write the image
>> to this allocated space in just a single curl_easy_perform call.
>How about those responses that don't have a content-length header? To make the
>solution solid, you can't rely on that header always being present. I would
>rather suggest you use a realloc concept like the getinmemory.c example that
>is totally independent from that header or not.
For now, I'm limiting my program to support only images with the
content-length header present. By the way, this is for a device with
limited memory and cpu capability.
>> So, my idea is this: Using a single curl_easy_perform call, I'll wait for
>> the content-length header in the CURLOPT_HEADERFUNCTION callback. Then when
>> this callback is called and I find the content-length header, I'll perform
>> the allocations, stop the retrieval of the headers, then start writing the
>> data retrieved.
>Why do it that complicated? You could instead let the header callback remain
>and just ignore the rest of the headers and you can have the write callback
>already set.
I can't set the write callback before the header callback gets called
because I haven't yet allocated the buffer that will hold the image at
that time. However, I did try to set the write callback earlier when I
was simulating this via PC and I noticed that the headers were also
being written to the buffer together with the image data. That's why
in the sample program above, after I get the content-length, I
immediately disable CURLOPT_HEADER so that when the write callback is
called only the image will be written.
>> Is it safe to call these curl_easy_setopts inside headerwrite?
>I'm not entirely sure, but I think it is.
Just want to ask, what makes you unsure? =)
My initial tests show the above hack works. But it's quite scary to
suddenly get an error with the hack in place during production of the
device.
Thanks,
Milmar