“app.webFetch is not a function” with safe_browser built from source


I have been setting up a private development network (for those who haven’t seen my numerous other support threads…) with the current status:
-5 vaults (had 8 but strange issues showed up until I cut them down to min_sec_size)
-Pre-built Peruse and WHM (using custom crust and routing files)
-Created public ID and website, uploaded sample info

When I attempt to build from source (pretty much followed the readme line by line with fresh checkout) then copyied in crust and routing files I am unable to browse my test safesites.

Why would this happen when building from source but not with the prebuilt browser binary?


This could again be down to uri registration. If that were failing the app never receives its authenticated state.

Does running xdg-open safe-auth://something trigger your built browser instance?


Ahh, xdg-open, the bane of my existence it seems.
I have attempted to reset everything by:
rm ~/.local/share/applications/maidsafe*
(manually clearing relevant mimeinfo.cache entries)

For whatever reason I end up with the following when launching the browser:
“xdg-open: file 'safe-bmv0lm1haw” …" does not exist" (this value is now listed under the authenticator entry in maidsafenet-ltd-safe-browser.desktop )

My system default browser is Chrome, but this also launches firefox (as well as running xdg-open safe-auth:test)

Sadly, this feels like it is becoming a Linux support issue rather than maidsafe, it is confusing as to why firefox is stepping in when it isn’t even the default browser nor defined in the .desktop entries.


I also get this error with the pre compiled version for OSX:

It works on version 0.6.0 that I still have. I tried to remove Library/Application Support/Safe Browser folder but with no result.


You may need to remove 0.6.0 @Yerontour. OSX and its url binding can be fickle.

(We’re looking at alternatives to this IPC process, as it is truuuly a pain :expressionless: )