Belgium
Write books in French, a blog on https://ploum.net and a gemlog on gemini://ploum.net
Comment by ~lioploum on ~lioploum/offpunk
On 24 mai 23 08:00, ~twijg wrote:
You are right, that is the simpler solution. This comment was maybe on the wrong todo but this is were I had the idea. I think it could still be a solution for 1* and 6* status codes. In the symlink solution, what would we do when the page it redirects to is not reachable? Save both for syncing and not create the first file?
That’s a very good question. Maybe, the easier solution is to make dumb redirection. We create the redirection then we create the error in the target, exactly like it would have been done if accessing the target URL directly.
I don’t see any drawback with this approach. (I may miss something, of course)
Comment by ~lioploum on ~lioploum/offpunk
I believe it could be a lot easier: "simply" make a soft link when a direction is redirected.
So gemini/bbs.geminispace.org/s/Gemini/12578 should be a simple soft link to gemini/bbs.geminispace.org/s/Geminispace/12578
This would also allows to follow the redirection while offline and display the good URL/cache
What do you think?
Comment by ~lioploum on ~lioploum/offpunk
Error has evolved : /home/ploum/dev/offpunk/netcache.py:471: CryptographyDeprecationWarning: The parsed certificate contains a NULL parameter value in its signature algorithm parameters. This is invalid and will be rejected in a future version of cryptography. If this certificate was created via Java, please upgrade to JDK16+ or the latest JDK11 once a fix is issued. If this certificate was created in some other fashion please report the issue to the cryptography issue tracker. See https://github.com/pyca/cryptography/issues/8996 for more details. c = x509.load_der_x509_certificate(cert, _BACKEND) /home/ploum/dev/offpunk/netcache.py:481: UserWarning: Country names should be two characters, but the attribute is 1 characters in length. common_name = c.subject.get_attributes_for_oid(x509.oid.NameOID.COMMON_NAME)
Should we ignore those warnings? Is this a bug in offpunk or in the certificate itself?
Comment by ~lioploum on ~lioploum/offpunk
Nice catch, that’s a tricky one.
Ticket created by ~lioploum on ~lioploum/offpunk
Redirects should be passed as argument to every netcache calls, including "is_cache_valid" to avoid many calls to refresh an URL (given that the original url is never cached)
Comment by ~lioploum on ~lioploum/offpunk
REPORTED RESOLVED FIXED
Am 15.02.24 um 22:01 schrieb ~lioploum:
gemini:// urls have still a problem, see gemini://[2001:67c:13c::7a2:de]/ or
Pushed a fix. This is now working.
Thank you a lot, it works!
That address is not working in other client. It doesn’t seem to be netcache/offpunk fault.
It's just returning "no proxying", that's ok.
Comment by ~lioploum on ~lioploum/offpunk
gemini:// urls have still a problem, see gemini://[2001:67c:13c::7a2:de]/ or
Pushed a fix. This is now working.
That address is not working in other client. It doesn’t seem to be netcache/offpunk fault.
Am 15.02.24 um 15:18 schrieb ~lioploum:
On 24 fév 15 12:56, ~Frank Doepper wrote:
an IPv6 URL, e.g. http://[2001:67c:13c::7a2:de], is handled fine with netcache and opnk, but offpunk says http://[2001:67c:13c::7a2:de] is not a valid URL to go
I think there's a mis-interaction between fix_ipv6_url and looks_like_url in offutils.py.
I’ve pushed a fix to both functions to better recognize IPv6. It is not very pretty (any cleaning would be welcome) but I hope it works without introducing some new bugs.
Thank you!
gemini:// urls have still a problem, see gemini://[2001:67c:13c::7a2:de]/ or gemini://[2a02:2770:8:0:21a:4aff:fe20:b137]/
but this is not in offpunk/offutils but in netcache.
Comment by ~lioploum on ~lioploum/offpunk
On 24 fév 15 12:56, ~Frank Doepper wrote:
Hi,
an IPv6 URL, e.g. http://[2001:67c:13c::7a2:de], is handled fine with netcache and opnk, but offpunk says http://[2001:67c:13c::7a2:de] is not a valid URL to go
I think there's a mis-interaction between fix_ipv6_url and looks_like_url in offutils.py.
I’ve pushed a fix to both functions to better recognize IPv6. It is not very pretty (any cleaning would be welcome) but I hope it works without introducing some new bugs.