Add DNS forwarder to DNSServer #7237
Conversation
Reduced size of a chunk of HTML string. Added maxlength to HTML input fields for credentials.
Added print statements for monitoring size of chunks etc. Novice improvents to HTML and variable names.
to avoid large allocations to buffer the send.
…to pr-dns-forwarder
…to pr-dns-forwarder
Plus 1 for new complier catch!
|
Is it expensive in RAM when it is enabled ? Those questions because I wonder what is against enabling it by default, and its impact on sketch that are not using STA. |
|
The forwarder queue size is a power of 2. It is currently at 8 items, total size of 96 bytes. I don't think an STA configured, enters directly into the DNSForwarder resource requirements. As you indicated it isn't very useful outside of using NAPT. From time to time I see complaints about the slowly increasing of resources with each release, so I tried to make it resource flexible. Also as it is written, it is easy to increase or decrease the queue size. It is a simple constant in the .h. However, for me, 8 items seemed to work well with one host browsing. With a smaller value, I saw more dropped requests. |
Sorry, I wasn't thinking that all the way through. W/O STA there is no need for the forwarder. |
| @@ -0,0 +1,330 @@ | |||
| #ifndef WIFIHTTP_H_ | |||
There was a problem hiding this comment.
why not .ino? just to avoid confusion with static things in headers and conditional typos
configHead, configHeadMin, and etc.
(although, I am coming from the external language server experience, perhaps it is easier in the current arduino IDE environment to work with files like this)
There was a problem hiding this comment.
...and also, why not a single .ino entrypoint, with .cpp actually implementing things?
There was a problem hiding this comment.
...and also, why not a single .ino entrypoint, with .cpp actually implementing things?
I don't think I follow this point.
There was a problem hiding this comment.
...and also, why not a single .ino entrypoint, with .cpp actually implementing things?
I don't think I follow this point.
Only keeping NAPTCaptivePortal.ino for setup() & loop(), for the sole purpose of avoiding ide concatenation of .ino files. Just a consideration, as I noted I may be out-of-loop and this is actually easier to manage within IDE
removed typedef Changed run_CI_locally.sh to run ci/style_check.sh. Copied resulting style changes from /tmp to PR
libraries/DNSServer/examples/NAPTCaptivePortal/NAPTCaptivePortal.ino
Outdated
Show resolved
Hide resolved
libraries/DNSServer/examples/NAPTCaptivePortal/NAPTCaptivePortal.ino
Outdated
Show resolved
Hide resolved
| DNSReplyCode _errorReplyCode; | ||
| bool _forwarder; | ||
| unsigned char _resolvedIP[4]; |
There was a problem hiding this comment.
btw shouldn't this be IPAddress?
libraries/DNSServer/examples/NAPTCaptivePortal/NAPTCaptivePortal.ino
Outdated
Show resolved
Hide resolved
libraries/DNSServer/examples/NAPTCaptivePortal/NAPTCaptivePortal.ino
Outdated
Show resolved
Hide resolved
…al.ino Co-authored-by: Max Prokhorov <[email protected]>
…al.ino Co-authored-by: Max Prokhorov <[email protected]>
The key functions added are:
bool enableForwarder(const String &domainName=emptyString, const IPAddress &dns=uint32_t)0)If specified,
enableForwarderwill update thedomainNamethat is used to match DNS request to this AP's IP Address. A non-matching request will be forwarded to the DNS server specified bydns.Returns
trueon success.Returns
false,dnsis not set, orvoid disableForwarder(const String &domainName=emptyString, bool freeResources=false)disableForwarderwill stop forwarding DNS requests. If specified, updates thedomainNamethat is matched for returning this AP's IP Address.Optionally, resources used for the DNS forward function can be freed.