Re: [Bug-wget] Segmentation fault with current development version of wget

[Darshit Shah]

Hi Giuseppe,

while I agree that your patch makes the code more readable and it is
> good to apply it, I don't see how the previous version was broken.  What
> compiler have you used?
>

This occurs when compiling on gcc 4.8.0. The following output ensues:

Setting --method (method) to head
DEBUG output created by Wget 1.14.43-cf74-dirty on linux-gnu.

URI encoding = ‘UTF-8’
--2013-05-05 07:56:19--  http://www.example.com/
Resolving www.example.com (www.example.com)... 192.0.43.10,
2001:500:88:200::10
Caching www.example.com => 192.0.43.10 2001:500:88:200::10
Connecting to www.example.com (www.example.com)|192.0.43.10|:80...
connected.
Created socket 3.
Releasing 0x0000000001271410 (new refcount 1).

---request begin---
EAD / HTTP/1.1
User-Agent: Wget/1.14.43-cf74-dirty (linux-gnu)
Accept: */*
Host: www.example.com
Connection: Keep-Alive

See how the request has the first character missing? And since --post-data
is also handled internally through this, it breaks all scripts that use
wget to send POST headers.

While on the topic of compilation issues, the linker flag, D_FORTIFY_SOURCE
throws a couple of warnings. Would we be interested in looking into those
and sorting them out?


>
>
Good catch!  Maybe we should handle --method=HEAD as opt.spider?
>
That makes sense.

I also think some more explicit sanity checks are needed in the code. Like
sending data with a HEAD request makes absolutely no sense (does it?).