[Edbrowse-dev] Forever

[Edbrowse-dev] Forever

[Karl Dahlke]

[-- Attachment #1: Type: text/plain, Size: 1329 bytes --]

Many sites, including amazon, which we've been playing with, take a very long time to plow through eb$qs$start().
Time is roughly O^2, as per the size of the web page.
No doubt we could rewrite it to be more efficient, but most of it is third party software querySelectorAll, that I don't want to rewrite, or even muck with very much.
They could come out with a new version with bug fixes, and I really want to just snap it in place.
Is there any way to make things more efficient, without going into the guts of querySelectorAll?

So this website hangs forever, and you can't even ^c out of it.
https://sploid.gizmodo.com/this-giant-fireball-over-michigan-was-visible-from-six-1822143868
Put this on the list of must-fix sites.
Now that edbrowse is one process, such sites are more troublesome; you just have to quit, and lose whatever was going on in the other sessions.
Following db 5 I thought it was eb$qs$start again, it hangs for a very long time, but then it returns, and then we hang on another javascript, seemingly forever.
We will need to address performance across the board, it's getting frustrating.
If that work waits til the next version that's fine, and I do want to talk about cutting 3.7.2 soon,
but nobody is going to use edbrowse if it hangs for minutes, and sometimes forever.

Karl Dahlke

Re: [Edbrowse-dev] Forever

[Kevin Carhart]

I know.  You're right.  I will try to pursue this.  I have actively 
procrastinated performance because this often was the next thing that 
unfolded after it didn't work at all, like on NASA.  (Note on NASA, I 
think it has had a regression and returns "1 character" again.  I think 
this has to do with cloneNode copying handlers which is in a jquery 
"supports" test.  Except we implemented that, so maybe this isn't the cause.)

> So this website hangs forever, and you can't even ^c out of it.
> https://sploid.gizmodo.com/this-giant-fireball-over-michigan-was-visible-from-six-1822143868

I get a libcurl error on this.  SSL connect error in libcurl: 
error:1407742E:SSL routines:SSL23_GET_SERVER_HELLO:tlsv1 alert protocol 
version

But maybe I can upgrade my libcurl and take care of it.

You know that person who wrote on github and reordered the sections of the 
README?  Do you write to people like that and try to allure them into 
coding something?  "As long as you've done a lovely job improving the 
instructions, would you like to do some more.."