Re: New Pepr link type detection is too strict

On 6/26/05, Sergio Carvalho <sergio.carvalho@portugalmail.com> wrote: > Hi, > > I was changing the XML_RPC2 status to 'Call for votes' and was greeted > with errors in the link types for the tgz and phps links: "The required > link type and the URL do not match!". > > Pepr validation code is being too strict. The respective links are: > http://pear.sergiocarvalho.com/XML_RPC2 (tgz file) > and > http://pear.sergiocarvalho.com/source/XML_RPC2-0.0.3/XML/RPC2 > (phps) > > I can understand the check on the tgz file, as my webserver is returning > a text/plain mime-type (erroneusly). I'd prefer linking the the package > name only, so people can install with: > pear install http://pear.sergiocarvalho.com/XML_RPC2 > as there is no reason pear must use only tgz as the package format; even > if it does use tgz, the URL can and should be agnostic to the format. > The resource being pointed to is the XML_RPC2 package. > > What I can't understand is how linking to an index of phps files can be > considered wrong. The package is not composed of one single file. To > link correctly to the package source, I must link to the index of all > sources, not to one specific file. > I responded to the original thread about this but got no responses. I'll say it again here, the URL itself *should not* be checked. The filename is not important in the least. What's important is that the correct link types are selected. The URL can be checked to make sure it's a real URL (minimal regex) and that it doesn't return a 404 header, but that's it. Any filename / extension should be allowed, In the case where the URL type doesn't match the URL posted the QA group can step in. -- Justin Patrin