Sam if you’re afraid you’ll have to overlook it,
Besides you knew the job was dangerous when you took it.
Ba-CCAAAWWWW!!
6 Likes
Another thread where “Show Full Post” isn’t working, and is just telling us to GET THE BOING BOING NEWSLETTER
Is anyone looking into fixing this? @beschizza @codinghorror?
1 Like
As has been discussed at length, this uses the “readability” algorithm to scrape the HTML and try to guess what is the actual content versus ads and cruft. It is a holistic, best effort heuristic kind of thing since depending on the degree of crazy of the source HTML it may or may not be practically possible to extract the “right” content without hand coding stuff for each article or website.
If you want to help, break down the HTML structure of the posts that have problems and point out the structural differences between those and posts where it does work.
3 Likes
Okay.
This post shows nothing when you click Show Full Post on the BBS thread.
This one shows the text, but not the video. We’ll ignore the video for now, that’s a separate problem to this one.
They both have a <div> with class featuredimage that contains the video. After that there’s a <div> with the id of story. The main difference between the two articles is that the drummer article has very little content inside that div, with only nine plaintext words, a single word link, and an image. Looking at the smokestack article the story div contains an entire paragraph, complete with link.
So the issue seems to be that the readability algorithm is ignoring content it considers too short which can be a problem on a website that sometimes post a link, and image, and the words “Look at it!”
3 Likes
In other words it needs a 
detector.
3 Likes
What I think it really needs is that hand-coded solution for this website: include everything in the featuredimage div and include everything in the story div that’s not part of a sub-div.
Ok, so what would the value be of having that expand, if it has so little content? Perhaps readability is making the correct call? If the only thing in the post is a video…
The problem is that videos are still broken so as far as the BBS knows that little content is the only content. And even if the video thing ever gets fixed sometimes those few words contain the source or attribution of the thing being posted.
Edit: Perhaps instead of hand-coding rules for each website, the readability algorithm could be changed so that if pressing Show Full Post returns nothing then a slightly different or relaxed set of rules could be used.
4 Likes
That’s a good idea! I like it. When nothing comes back, fall back on a different strategy. For those “I just posted a video and 6 words” posts.
4 Likes
“For the second time!”
2 Likes
Here’s a new problem I’ve never run across before. Clicking on Show Full Post on
somehow strips the link out of the first sentence of the post, leaving the post as a frustrating & tantalizing description of what the reader could find if indeed they themselves could experience what Rob was talking about.
Er, clicking “Show Full Post” on that one 
seems fine to me? Am I missing something?
You missed that time on Monday morning that it didn’t work. Maybe @beschizza edited the post to fix it, or maybe it was a temporary glitch with Discourse, but at the time I wrote that there was no link in the Show Full Post version of the post while there was a link on the actual BoingBoing post.
1 Like
Oh suuuuuuuuuuuuuuuuure fiddingfrog, suuuure there was a problem. 
2 Likes
Going back to this readability problem…
This conversation:
only shows “GET THE BOING BOING NEWSLETTER” when you click on Show Full Post.
But the article:
clearly contains more (2 images, 16 plaintext words, 6 linked words). In fact this is the longest article I’ve seen readability fail on.
1 Like
So, do you get the newsletter?
#OF COURSE NOT
5 Likes
5 Likes
New record:
An embedded image and 54 plaintext words becomes “GET THE BOING BOING NEWSLETTER”.
3 Likes
been noticing this a lot, lately, too.
ETAT: w/r/t Israel B below, yes, I’m FF on OSX
1 Like
Problem comes and goes with any version of Safari (OSX/IOS) or Firefox on OSX.
1 Like
