Closed cgf1 closed 3 years ago
This should probably be the default, but with a setting for when you absolutely positively need something else.
I'll look at adding this soon, thanks
Sounds like the right solution. Thanks.
Given the description, I'm not sure the attached patch is the right move, as that's bit of code simply fetches the description of the spec from the WC3. This should only be done once on given installs, rather than on every instantiation of a Selenium::Client or fetch of a page.
If we had something happening per fetch, it would mean the selenium servers themselves are returning UTF8 which is not being handled correctly by the JSON decoder.
If this is what's going on with your client, there is a more elegant fix.
It would be helpful to get the right fix if I had the precise error messaging. If it's something like "wide character in"... during the HTML::Parser logic discerning the spec, then this is probably the answer.
That said, I don't encounter this myself on any of my test systems, so I am not sure this is the case / have no way of knowing.
Try out the above patch in 058284e, see if that works.
I tried to do things the "right" way all around.
Sorry that I haven't responded more quickly to this. The subject of this bug was the error I saw. I have only seen it on one site that I tested. I'll try your patch in the next couple of days.
Resolving, will push what I've got, having heard no opposition.
I got the error in the subject from a client site. The attached patch fixes it but I'm not sure if it is the right solution or not. p.txt
That assumes that every web site will be using utf8 encoding which I'm not sure is correct.