Hacker News Reader

Comment by FussyZeus | Hacker News Reader

Also as a quick aside, they mention that AutoTrader would've had to fix "hundreds of thousands of pages" and I'm calling bullshit. No site of that magnitude is using straight up HTML pages. Even if it's a jekyll thing where you'd have to change a bunch of templates, I'd be willing to bet the vast majority of the required changes could be done in a few seconds with any decent text editor.

View on HN

I would not be surprised if they are telling the truth here.

I worked on an early version of a big woodworking website and it was over six thousand static html pages.

The group that acquired it needed to show progress asap, which meant simply updating the header to their new logo and footer nav.

I had to write about a dozen sequenced find and replace routines with regex to do even basic changes. In some cases groups of pages used subtly different html it was a real mess.

It took a long time for them to get all the pages replicated in a templating system to avoid seo loss.

The thing is though that's a bad system, whether EasyList is prompting the changes, or the SEO, etc. and it feels like they're trying to pin "we had to fix thousands of pages" on EasyList as though their shit workflow is EasyLists' fault.

The thing is is that AutoTrader is a website that advertises cars, they don't even have any other advertising on their site. And because EasyList doesn't like a word in their URL, the website breaks through no fault of their own. If the owner of the website is doing nothing wrong, the onus is on the owner of the blocking technology to come up with a system that works correctly.

I just visited the AutoTrader site (from the UK), and uBlock Origin blocked 34 requests related to ads and tracking. I recall in the past having been surprised at the number of extra requests being made via their site. They don't serve only vehicle listings.

It would be rather ironic if the post you replied to only thought they ran no ads because they forgot they ran an ad blocker.

Ooops, that's my misread of the forum posting from the AutoTrader team. They have 2 sites, one public, the other is a portal for car dealers to post cars in which they stated does not serve ads. The problem they were having though is that the ad blocker was affecting popup menus within their site.

https://forums.lanik.us/viewtopic.php?f=64&t=43148&sid=e0edf...

In that light, I agree with your original comment, though it's amusingly ironic that their portal site was negatively impacted by mitigation measures against the kind of things their public site engages in.

I've been working in big tech companies for too long. When I got to that part, I thought to myself, "Wait, they think hundreds of thousands is a lot of content? That should be fixable over someone's lunch break!"

What big tech company do you work at that can edit other people's websites to link to renamed URLs? I am sure that "mv /www-data/adverts /www-data/foobar" is a trivial operation. Making links work (adblock will block the 302 you skillfully put in place) is harder.

Ahh--the lunch break comment assumed that this was a local change to hundreds of thousands of pages, so disregard that. I was commenting more on the notion that hundreds of thousands of files/urls/database records is a lot. It sounds like a lot to the average reader.

You would be surprised. Many years ago I worked on a CMS for a large household-name client. Their entire website was made up of individual HTML pages. No shared header, footer, or anything; if you wanted to update one of those, you would be updating it thousands of times.

My task was to move all of that into a very minimal CMS. All the content would be in flat files, and the page itself would load the header, footer, content, etc.

Because every page was different, but mostly similar, I wrote an Emacs extension that let you identify the content areas (it tried some heuristics and you could guide it) and it would do the extraction and emit the flat content file and the replacement JSP file.

What you have to realize is that many people have systems that are designed for 1 "thing" and have since "scaled" that up 1000x. People still treat their website as a directory of arbitrary HTML. People manage 1000 servers like they managed their one and only very first server. It is the natural way of growth. Engineering is figuring out when that model doesn't apply anymore, and changing it. Engineering is in short supply, however, leading to stories like mine above or the "thousands of pages" blocked by EasyList.

I used to work at a commercial website host where the primary product they sold was "promo pages for your xyz business and abc products" and each and every single page was manually updated down to the head/footer and statically served. They had "templates" in a notepad, and you copy and paste it out and edit as needed. Which means that after you make the initial creation, any subsequent edit (including things like headers, meta tags, etc) needs to be done by hand. This product was still being sold very very recently, and they have a lot of clients (like in the hundreds of millions of dollars of revenue)

I wouldn't be at all surprised if AutoTrader used manual editing for everything, with no access to scriptability.

Not everyone is using jekyll or any modern stacks ... the echo chamber is strong here. I guarantee you 65% of the Internet is still being edited by hand.

This is all just speculation, but this attempted hit-piece on EasyList appears to be slowly pushing AutoTrader in front of the approaching bus.

It’s perfectly possible.

I once worked on a site that had over 25k pages, all hand coded HTML with Microsoft Frontpage. And all in the root folder.

Not even the owner knew how many pages there actually were.

Reply on news.ycombinator.com