A lot of packages to build Linux distributions for the embedded world relies on "wget".
Typically they are based on a Makefile and a configuration file included by the Makefile.
 
If a package is to be built, then the package is downloaded to a directory somewhere.
The makefile will try to extract the package from this directory first
and if it does not exist it will try to download the package from an Internet site
using wget.
 
It is a quite common problem that the file does not exist, because
a new version is available and the original file has been moved to another location.
 
wget would be significantly improved if it would try one or more alternate locations for the package.
 
When wget reads the configuration file(s) ~/.wgetrc.xml
this could contain information about alternate sites.
 
Ideally, these files should contain information about
* list of files containing associations between package names and
    where they can be found, either on ftp, http sites or on the local disk
* Web sites from where new lists of associations can be downloaded.
 
In addition there should be the files containing the associations...
 
In order to avoid having to rewrite a lot of scripts there
should ideally not be a switch in the wget "download" command which indicates this.
It is better if the configuration is retrieved and the fact it is there
is enough for wget to try.
 
Something like
 
would of course work if someone set up such a site, but a local solution is better.
 
 
-----
It would not be a bad idea if wget could report all files at the failing site
which are similar to the one requested.
 
"wget --<switch> <site>/<package>-*.tar.gz"
should give a list of all packages that look like <package>-*.tar.gz/tar.bz2
wget --spider somehow does this, but results in a lot of extra unneccessary info.
I just like to get the filenames.
 
 
 
 
I do not subscribe to the wget mailing list so please reply to "ulf at atmel dot com"
 
 
 
 
Best Regards,
Ulf Samuelsson

Reply via email to