Consider DNS discovery about tent.io HOT 9 CLOSED

+1

from tent.io.

+1

from tent.io.

Strongly disagree. In fact, I'd suggest that in an ideal world you shouldn't need to have ownership of a domain at all, and install a Tent server in a subdirectory on a shared host. Adding discovery via DNS TXT records is highly technical and beyond the realm of (perceived) possibilities for many users.

from tent.io.

I disagree as well. I support adding this option, but not that it becomes the only requirement for Tent discovery, replacing HTTP header and HTML .

from tent.io.

DNS requires domain control, which the current header-based discovery doesn't. It also is very complicated for users to setup, and wouldn't allow Tent entities to use subdirectories (as @benwerd points out).

from tent.io.

@benwerd installing a module on a shared host typically means clicking a thing in a web-based control panel. The TXT component could easily be handled by the host's component installer, abstracted away from the user. In the case that you mean a shared host where the user is performing CLI installation, then I contend it's axiomatic that the installation must be at least as or more complicated than adding a TXT record, which is one step. There are other issues with the promise of using shared hosting around requiring SSL certificates.

@kiliancs in saying it should be the "only requirement", I meant it should be ordered with the highest priority in the discovery sort. I.e. DNS supersedes Headers, Headers supersede HTML content.

@titanous DNS does require domain control, but header-based discovery requires a HTTP server. Every domain name registrar provides DNS hosting. Web hosting, especially at the level where headers can be manipulated, is more onerous. For all Tent server installations, adding a TXT record is at least as easier if not easier because adding a DNS record cannot be more than one step. Subdirectories are good in theory, but there are issues surrounding SSL certificates in multi-tennant setups that present more complexity than a DNS entry. Finally, nothing prevents the TXT record from being formatted such that a) there are multiple entries per domain, and b) subdirectories are supported on the entry.

My proposal could've been more clear: I think the headers/HTML content discovery are good ideas, and should remain, but believe DNS discovery would dramatically simplify 3rd party hosting. As it stands without DNS discovery, one's Tent stack is tightly coupled with one's web stack. If a users does not desire a web presence, they cannot use Tent.

Email and WWW, protocols Tent aims to imitate, use DNS discovery. Every web user has an email address--they're all somehow able to navigate DNS records despite DNS records being "highly technical and beyond [their] realm of possibilities" @benwerd, mostly through 3rd party hosting.

If Tent doesn't enable (and prioritize) DNS discovery, then it's not a first class HTTP protocol like Email and WWW. With only Header and HTML content discovery, Tent is a web server module. A web server module is not compelling enough to win.

from tent.io.

If Tent doesn't enable (and prioritize) DNS discovery, then it's not a first class HTTP protocol like Email and WWW. With only Header and HTML content discovery, Tent is a web server module. A web server module is not compelling enough to win.

I absolutely agree (I've been lurking watching this thread) I'm a little surprised that after speaking on the homepage about how you want Tent to act like Email and the web, then rely on one of those stacks to propagate.

DNS is a wise move, and in the end it shouldn't be a massive topic, it's yet-another way to do discovery, it should be a "core" way, as I could absolutely imagine an off-shoot of Tent to support it unless the core spec defines that.

from tent.io.

Tent entities to use subdirectories

Is another interesting point, and isn't precluded by dns discovery (the TXT record could be formatted such that it enumerates the known-tents on the domain, even if they are directories/other physical servers) - but perhaps the discussion needs to settle a little and for us to identify what the sole objective is.

(Seems like it's absolutely flexible hosting options, in which case one can't ignore DNS as a discovery mechanism, it works for Apple's bonjour, for example.)

from tent.io.

@leehambley I believe it also works for Apple's Back to My Mac

from tent.io.