Re: [Bug-gnubg] Problem with source snapshots - Browser Downloads are in ASCII?

[Michael Petch]

On 08/02/2011 5:07 AM, Michael Petch wrote:
> Seems to be browser download related. The file downloaded by wget (for 
> me at least) has no issues and ls -l shows:
>
> -rw-r--r-- 1 root root 14199163 2011-02-06 20:50 
> gnubg-source-SNAPSHOT-20110207.tar.gz
>
> The file downloaded by firefox is smaller for me (and exhibits the 
> same behavior you see). ls -l shows:
>
> -rw-r--r-- 1 mpetch mpetch 14194658 2011-02-08 04:54 
> /home/mpetch/Desktop/gnubg-source-SNAPSHOT-20110207.tar.gz
>
> As well Christian's method of using gunzip first works.

I discovered what was going on (but not why). I should have realized 
that one of the ways a download would be larger (and not work) is if the 
binary file was transferred as ASCII and not binary. Firefox is 
downloading with ASCII on so we are getting extra characters because of 
the CR/CRLF translation that gets done.

gunzip seems to work because it is smart enough to realize that the file 
is corrupt because of CRLF translation issues, so it fixes it on the 
fly. tar's decompression isn't that sophisticated so it throws an error.

When I have time I'll take a look at the headers returned by Firefox on 
an http download. There must be something in there that is causing the 
tar.gz to be downloaded as ASCII rather than binary.

I seem to recall other posts in the recent past about browser downloads 
appearing corrupt. Probably all related.