[Edbrowse-dev] other bugs

[Edbrowse-dev] other bugs

[Karl Dahlke]

[-- Attachment #1: Type: text/plain, Size: 1399 bytes --]

The problem with using third party software is we sometimes find their  bugs, and then hope they fix them, or we work around them.
acid test hunts for corner cases so not surprising it uncovers bugs.
I already wrote about the tidy bug, not giving us the empty nodes, which test 0 expects to be there, and so I guess every other browser puts them there.
I sent this question to Geoff.
But turning to css, there are more problems.
The parser that Kevin found for me doesn't parse comments corectly.
If a comment contains a selector it is treated as a selector, not a comment, and it's all screwed up from there.
I reported this to jotform, no response yet.
Ok this one seems like an easy work around.
I don't need the comments preserved for any reason, so just strip them out with a preprocessing regexp.

css_string = css_string.replace(/\/\*(.|\s)*?\*\//g, '');

Yeah I know, it looks like line noise in the modem, but it's right, I'm pretty sure, and works on the modest test strings that I give it,
but it doesn't work on the style block that is part of acid test 0.
RangeError: regexp executor recursion limit
I used the nongreedy modifier to the wild card, so there really shouldn't be any recursion at all, so I don't get it.
I sent this to one of the duktape developers.

With all these issues hanging, I'm rather in a holding pattern as per the acid tests.

Karl Dahlke

Re: [Edbrowse-dev] other bugs

[Kevin Carhart]

Thank you for discovering these.  Too bad but I guess like you said...

> acid test hunts for corner cases so not surprising it uncovers bugs.

Exactly... it's going to exercise a lot since that's what it's built to 
do.  One piece of good news is that a few days ago when we added something 
and then I said not "it works," but "we now have the pleasure of having 
exposed the next problem.  It is frustrating but actually a good thing," I 
temporarily stayed mum on what it was we had exposed.  What we had exposed 
was that for me at least, edbrowse was now hanging for an unrelated reason, 
way off in test 43.  Should not be misinterpreted as a failure even though 
the outward manifestation is "it's hanging, this is bad."

So the punchline is that as of the last couple days with the CSS code, 
edbrowse no longer hangs.  I don't know the cause because test 43 is way 
off, but the cessation of hanging is nice.

Was it a helpful format to send cloneable .git links for the two projects? 
Because, in case it was, here is the git link for the javascript DOM that 
I have mentioned:
git clone https://github.com/thatcher/env-js.git

If you clone it, go to src.  And then there is a css folder.  Maybe there 
is something in there worth cribbing or referring to that you (KD or 
others) would like to know about.

Kevin







> I already wrote about the tidy bug, not giving us the empty nodes, which test 0 expects to be there, and so I guess every other browser puts them there.
> I sent this question to Geoff.
> But turning to css, there are more problems.
> The parser that Kevin found for me doesn't parse comments corectly.
> If a comment contains a selector it is treated as a selector, not a comment, and it's all screwed up from there.
> I reported this to jotform, no response yet.
> Ok this one seems like an easy work around.
> I don't need the comments preserved for any reason, so just strip them out with a preprocessing regexp.
>
> css_string = css_string.replace(/\/\*(.|\s)*?\*\//g, '');
>
> Yeah I know, it looks like line noise in the modem, but it's right, I'm pretty sure, and works on the modest test strings that I give it,
> but it doesn't work on the style block that is part of acid test 0.
> RangeError: regexp executor recursion limit
> I used the nongreedy modifier to the wild card, so there really shouldn't be any recursion at all, so I don't get it.
> I sent this to one of the duktape developers.
>
> With all these issues hanging, I'm rather in a holding pattern as per the acid tests.
>
> Karl Dahlke
>

--------
Kevin Carhart * 415 225 5306 * The Ten Ninety Nihilists