Adds support for the new uri schemes introduced by cashaddr.
Details
- Reviewers
deadalnix schancel kyuupichan - Group Reviewers
Restricted Project - Commits
- rSTAGINGa38bff37c337: Introduce bitcoinUriScheme
rABCa38bff37c337: Introduce bitcoinUriScheme
Added tests
Diff Detail
- Repository
- rABC Bitcoin ABC
- Branch
- 8-12-uriScheme-only
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 1351 Build 1351: arc lint + arc unit
Event Timeline
I was under the impression that we would be gracefully degrading in the case of base58 addresses. This looks like there's some toggle flag that's required for the user?
src/qt/sendcoinsentry.cpp | ||
---|---|---|
39 | It seemed like you were threading the config through the GUI. Should we pass it in here also? |
src/qt/sendcoinsentry.cpp | ||
---|---|---|
39 | You're right. How about doing that in a separate diff after this one? |
src/qt/paymentserver.cpp | ||
---|---|---|
205 | This sinkhole all the configs. It looks like what you want is a ProxyConfig class that wrap an existing config and forward all its method to it, and then override a method from it. Or maybe a way to clone a config and then change some of its parameter. Either way, this is not good design in its current shape. | |
230 | Really, cfg is not significantly shorter than config. Also, what tells us we are supposed to be on main net here ? | |
411 | dito |
src/qt/paymentserver.cpp | ||
---|---|---|
205 | There is no existing config to wrap at this point, so I’m not sure what you mean. Will you be working on new this config architecture (ProxyConfig)? Should this wait for it? | |
230 |
This change does nothing but remove the URI_SCHEME constant. The functionality (or lack of) is the same. The fix to support multiple schemes here is in D582. |
src/qt/paymentserver.cpp | ||
---|---|---|
205 | The problem here is that thing thing is clobbering all the other configs, not just the use of cashaddr. The only reason this work is because you know that the code down the road is only using that, but this is breaking the whole abstraction. | |
230 | No, it is not the same. The network in the old system is determined by a prefix coming before the hash in the base58 encoded data. In this case it comes from the prefix. This is breaking properly functioning code. It sounds like D583 should come before this one. I guess this is ok if you this exclusively to fetch the prefix, but in this case, I don't see how it is an improvement over using a constant, and if this is used for anything else, this is quite dangerous. |
src/qt/paymentserver.cpp | ||
---|---|---|
205 |
This is why I used Params.CashAddrPrefix() to get the scheme instead of a dummy object in my previous diff revision (rejected because it didn't use bitcoinURIScheme()). How about creating a second bitcoinURIScheme function with signature (const CChainParams&, bool useCashAddr)? Then we don't need a dummy object. The (const Config&) one can still exist for convenience. |
src/qt/paymentserver.cpp | ||
---|---|---|
205 | I'm good with that. |