-
Notifications
You must be signed in to change notification settings - Fork 1.1k
Conversation
add comment block for nonfunctional hosts
This is a really useful tool to create simple rules, I think we should also recommend this in our guides. We may also give a hint in this script to manually search for more subdomains. Or even providing the search link? I decided to use the letters from the superscript minuscule Unicode table for common errors. This hopefully brings more consistent and order in our comments. |
@J0WI I think it's great you did this work and I definitely like having a template top comment. I don't like the superscript pattern because it's harder to read, and Unicode can be annoying. I'd much rather just use When you specifically list categories, contributors who encounter a problem that's not listed like Should we ask contributors who submit a new ruleset with no problematic subdomains to delete the top comment? Also, when I make new rulesets, I just copy an existing ruleset and edit it. I haven't used |
Personally I prefer Unicode symbols for readability. We also already have quite a few rules using them:
The initial idea of the comments template was, that I was not able to type these characters on my keyboard and it was annoying to search them. So I wasn't using them.
The best practice for mixed content or cert chain issues would be to use a platform.
I thought this is self-explaining, but we could easily do that. |
I know we have a bunch of rulesets with this Unicode pattern, I've seen them and don't like it. The Unicode is just plain hard to read for my eyeballs, even with a large monitor. Also I feel that almost anyone would agree that
Is easier to write and review than:
Also I really don't want to have to walk a contributor through using Unicode footnotes to comment something like |
The sort-by-category solution is nice for small rulesets, but when you have many commented subdomains you have to sort each category separately. When you have all comments of a host in a list, you can easy add new ones and then just sort the whole list. Therefore it's also easier to find a host in the list, because they are only sorted in alphabetical order and not spitted into categories. Maybe you have more than one domain in one ruleset, then you need to sort it in three times in the first example and only two times in the second:
vs
In the category list you also can't note multiple issues, e.g.:
(Means bad.example.com serves a mismatching cert, if you accept that it will redirect the wrong location and to http) |
I think there's a middle-ground here. I agree with the readability of non-superscripted characters, as well as the niceties of having one master list and the ability to tag multiple times for a given host. These are not mutually exclusive. Take the following:
Look good? |
These are some more characters to type, but I'm also fine with that. |
If we're going to use annotations then I agree regular ASCII is better than Unicode superscripts. Does I think the exhaustive list of categories ought to be this, with whatever names we find appropriate:
That's it. Possibly Also see the discussion here. |
@jeremyn I think we should settle on some set of standard categories and note them in https://github.com/EFForg/https-everywhere/blob/master/ruleset-style.md. But I don't think we should prematurely limit ourselves to a predefined set and prevent ourselves from adding more. Who knows what situations will come up. For instance, as SHA-1 certs become deprecated, if the StartSSL SHA-1 heisenbug presents a problem for many of the subdomains on a specific ruleset, we'll want to note that. So I don't see any reason not to create a new category for it. |
If we're going to pre-list categories in the trivial rule template, the list of categories should minimal, obvious, and complete. |
I plan to add a CONTRIBUTING.md file when this one is done. |
With |
I just updated this without Unicode chars. |
lgtm! |
improve make-trivial-rule duplicate heuristic
add comment block for nonfunctional hosts