With a plugin, you have a choice … so take your pick of whichever system you’d want.

‘better’ is subjective. To each their own. For me I don’t know what system I might prefer yet. Will wait and see how things progress.

1 Like
  1. From rfc safe://<subName>.<publicName>/<path>
  2. Why not safe://<publicName>.<subName>/<path>

For example

  1. safe://
  2. safe://london.westminster/bigben/clockroom

for 1.:<subName>.<publicName> hierarchy right to left, but <path> hierarchy left to right. That is inconsistent.
Also for automatic tab completion in commandline it seems more practical for me to start with publicName.
There are probably better reasons to keep 1., but it doesn’t hurt to double check that.

1 Like

I think using clearnet terminology is confusing at least as much as helping. So it may be better to use SAFE terminology and define that.

I think the SAFE terms are public name and service (but might be wrong about the latter), as in: safe://<service>.<pubname>

The <service>. portion can be left off a URI in which case it is treated as www. and is used to look up what ‘service’ is available at that URI on the <pubname>

Services might be a website, email inbox etc., and new services can be added later. I added one called LDP to support Solid apps in my demo (although a bug, I think in how the browser interpreted this, meant I had to undo it and use ‘www’ instead).

What comes after that is moot. We think of it as a path because that’s what it is on clearnet, but it can be anything and what and how it is used will depend on the service. A particular client may be needed for a particular service, so the browser will know about some standard services, but not new ones - those would only make sense to the apps which know how to interpret them.

I hope that clarifies rather than muddies this discussion, and I hope I’m right in calling it a ‘service’!

Correction: Having been digging into the relevant code I see that the name for ‘service’ is, I now remember ‘subname’, such is used to look up the service in the ‘subnames’ container (previously services MD). Sorry for muddying the waters! :blush:

I wish I had a better memory. I’m pretty sure I argued for subName to replace service in the context of a SAFE address because it is different from the service itself. E.g. safe://thing.happybeing could be web, or email, or any other service. How could I forget? Doh!



If even @happybeing is unsure of what to call the different parts currently required by at least the Web Hosting Manager, then there certainly is room for improvement. What is the state of the WHM anyway? Is it still the best place for a new user to start if he wants to create some simple HTML pages with links between them?

If one creates a new phenomenon, one also has to create new terms. Natural language doesn’t always work that way, but when it comes to IT, the language used isn’t really “natural”. “Terms” aren’t just any words, but well-defined labels. And for the sake of clarity, it’s better not to reuse existing words and introduce unnecessary ambiguity. Why use “carriage” for “automobile” when inventing one? Natural language words like “auto” or “bil” come later when the phenomenon enters public consciousness and there is less need to be exact. I’m no inventor, but I’m used to reading (and thinking) from left to right. The automatic tab completion mentioned by @draw is also a good point.

Honestly, I don’t think this must be decided as per requirements from a technical perspective. I think it’s more about what is easy for the mainstream consumer. If I’ve never heard of the safenetwork and I’m going to use it, I want it to be intuitive and recognisable.

I think for all humans with internet, the way to go to a website is to do www.xxxx.ext. So in my opinion we should stick as close to that as feasible to make the transition from the old web to the new as easy as possible.

Just skimmed this so far (catching up after a week away). @TylerAbeoJordan incase you didn’t see it there’s an RFC for public name resolution as @draw linked to above.

Will come back and read this thread in detail later :+1: :bowing_man: interesting ideas!


Also in the RFC (see post above from joshuef): ‘subName’ instead of service or sub domain is used. But in the XOR-URLs RFC I see service being used.
I should read these RFC’s properly again later, but it could be that there is not much difference in behaviour/functionality between: safe://b.a/c and safe://a/b/c. Because if there is no or little difference what is the added value to go from safe://b.a/c to safe://a.b/c instead of the already possible safe://a/b/c.

1 Like

I agree with @happybeing we are not duplicating the current internet and so “service” is 100 times better than subname which implies the wrong thing


I wasn’t really advocating for ‘service’ and am vague about the history of this SAFE terminology. I was saying that I think it is clearer of we use the current SAFE terms for these things because they are different to the clearnet terms.

I’m not sure if service is the best term and have I think argued against it in the past!

Plus, I’m not sure if Maidsafe plan to maintain this services model, so it might well fall away. I quite liked it as a way for developers to expand SAFE services to support new things, but I didn’t get much feedback on how this was expected to pan out when trying to understand it and apply it myself so I think Maidsafe might still not be clear on this.


If I remember rightly, what was proposed as services fits neatly into using RDF in the resolvable maps. So you can describe a subname (or what have you), via RDF and so some apps may choose one of the entries in there (‘service’) over others depending on schema (Inbox, over FilesMap eg).


The SN is already doing that:

Also what are the problems you’re referring to?

Of course it’s a dns thing, but you can make multiple domains point to the same server, then the server needs to deal with them.

I was WRONG: (see @neo’s post below correcting me)

why aren’t you just striking it out then? (~~foo~~ => foo)

But isn’t that the way domain servers are been organized, the root dns server (“.”) points to a TLD server (“com.”), the TLD server points to the sites server (“”). Thus the organization managing the root server(s) “own” the entries to the TLD servers?

hmm, yeah, but the “categories” are also a nice feature like “” or “”. But we could resolve them as a flat name, so you would own the id “” (or “xxx.pornsite”, which ever you like more), no ownership hierarchies needed.

please no… just remove file extensions, there are much better systems to define the type of a file (eg. MIME).

i would go away with the “service” portion in names, and just use a context sensitive resolution in software. So if you go to an id on the web browser it will try to resolve the “web”/“www”/(what ever you wanna call it; maybe just a well-known UUID, comparable to GPT, it is using UUIDs for the partition types) portion implicitly. If you’re opening an id in a video app (youtube) it will fetch the videos, a messaging app will open a send form with the id as the recipient, …

1 Like

No because google own “” and apple own “

But in safe every subdirectory under com belong to the one ID and that ID has been reserved the name “com”

So if your name subname/service then it is still the one ID that has the name “com” no mater what “xyz” is be it apple or google. Not separate for each subname/service.

This is another reason why the “dot” should only be for service to remove any confusion.


Yeah, and someone else owns “com.” and “.”. The SN equivalent to that could do just the same, the “” domain entry points to a MD, that is owned by apple.

How is this different to the current dns?

Because in SAFE only one ID can be for

SAFE has a single ID key pair with the NRS record for “com” And all those subdomains of “com” would all fall back to the ID “com”

DNS has a separate name record for every one of those million subdomains for “com” and the registrar does not own them. Under safe only the person with the key pair for “com” has ANY ability to work with all those subnames/services


yeah, and that dude who’s owning the “com” id, would add links to “”, “” and so on.

Yeah, but the registrar owns the entry to the subdomain, he can even redirect it to an other server.

It will be a fight to get the ‘com’ public name then.


Yea and its not hierarchical DNS where subdomains have control over their subdomains. SAFE is a flat naming system

The owner of “com” cannot give google control over since google is only a subname (really service) of the ID that owns the “com” AD

Unless you want the stupidity of giving the keys to the account with the ID that owns “com” to every subname user and hope they play fair. They have to store their web page files under that account. And links would go to the Name that they link to, not stay under “com”

SAFE is a flat naming system NOT a hierarchical domain/subdomain system

1 Like

I’m pretty sure I’m not in control of the DNS servers pointing to my server… I’m fully relying on 3rd parties to make sure things run smoothly and wouldn’t recognise if visitors would be redirected to a different server :roll_eyes:

So renting out “sub names” is not something that would be a no go in our world…

You should be in control of your DNS records

In any case teh DNS system relies on hierarchical structure and the design is for various entities to maintain the DNS records be they the actual owner of the domains or 3rd party.

And more importantly when you go to a then is where the requests are going and the DNS system wades through DNS servers to come up with the IP address.

SAFE is just a name and if you have “com” then anything “com” is unavailable to anyone else. If you try setting up a service to link to other names then the browser will have that name (and subdirs) showing in the address bar. This is because you went to another site.


  • by linking (see above)
  • By storing their web pages in your account. Well trust issues there and ability to update website
  • By giving to your customers your account keys so they can update their website and use their and that portion of your account’s file system?

What I don’t understand why are people are trying to go back to a hierarchical system that has a lot of issues and replicate those in SAFE???


It’s not like one couldn’t set up a server attached through signalling that can manage the links

Would be a bit stupid but that doesn’t mean nobody would want to do it :man_shrugging: