On Fri, 26 Feb 2016, bch wrote:
What I suspect that original code was going was parsing the content from a gopher server (and in fact, taking some shortcuts that may have happened to work for original use-case). The actual URI is the second component of an appropriate gopher line, and to be used unadulterated. What I think the previous code was supposed to do is take a raw line from a gopher server and eliminate it's "status", and use the user-visible component of the returned content as the URI. That sort of parsing is better left for a higher-level client. This patch removes the test of path length and stripping characters, but preserves tab handling (arguably still removable) and escaping.
So our existing gopher tests aren't good enough or are they plain wrong? Have you looked at adding a test that verifies what your patch fixes?
-- / daniel.haxx.se ------------------------------------------------------------------- List admin: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.haxx.se/mail/etiquette.html
