cURL / Mailing Lists / curl-library / Single Mail

curl-library

Re: How bad is "Empty reply from server" ?

From: jérôme Muffat-Méridol <jmuffat_at_webphotomag.com>
Date: Fri, 15 Feb 2008 16:29:17 +0100

On Fri, Feb 15, 2008 at 11:44 AM, Daniel Stenberg <daniel_at_haxx.se> wrote:
> On Fri, 15 Feb 2008, jérôme Muffat-Méridol wrote:
>
> > yes, premature is true.
>
> PLEASE don't top-post, now we just lost the entire context of this mail and
> readers of this mail will have a hard time understanding what we're talking
> about.
>
>
> > Now, looking in detail, this patch doesn't accomplish anything:
>
> Eh?
>
>
> > 1) It avoids calling failf, which only affects stderr
>
> First, that's not the only thing it avoids since the block both calls failf()
> and returns CURLE_GOT_NOTHING.
>
> Then, failf() does not only affect stderr.
>
>
> > 2) it returns CURLE_GOT_NOTHING instead of CURLE_OK.
>
> Why?
>
> --
>
>
> Commercial curl and libcurl Technical Support: http://haxx.se/curl.html

Sorry, for my bad usage of this mailing list. You'd rather I put my
message at the bottom, right?

Here is what I meant:
I have done the changes you suggested and it didn't do anything new.
Tracing the code, I have come to the conclusion that the change didn't
have any significant effect:

- first effect: it avoids calling failf(data, "Empty reply from
server"); which (at least here) simply avoids outputting a message
to stderr / debug callback.
- second effect: returns CURLE_OK instead of CURLE_GOT_NOTHING . But
tracing further, that return value isn't used down the line at all.

So, adding this test on !premature isn't really achieving anything.
That's ok, guesses are just that: guesses.

I have made a separate small program that requests one file, cancels
and requests another. And as Murphy would have it, it isn't showing
the wrong behaviour yet (I'm still trying to get it to failf(data,
"Empty reply from server"); )
Received on 2008-02-15