Skip to content

Mobile-friendly view for journey

Announcements
6 3 1.4k 1

Related Topics
  • Ch..ch..ch..ch..changes!

    Announcements themes layout
    16
    2
    15 Votes
    16 Posts
    3k Views
    @phenomlab of course, to be recognised is fantastic. @phenomlab said in Ch..ch..ch..ch..changes!: Sadly, no. Web crawlers and scrapers are often JS based and read text only, so styles don’t have any bearing. I’ve read mixed things about this, but no that does make sense, it was something I read a many years back when using Wordpress.
  • ANNOUCEMENT: New NTFY Server

    Announcements ntfy push alerts
    9
    1
    7 Votes
    9 Posts
    2k Views
    @crazycells that’s as good a test as any
  • 36 Votes
    43 Posts
    11k Views
    I was experiencing 500 (Internal Server Error) responses from the proxy, visible in the browser console: GET https://proxy.xxx-xxx.net/ogproxy?url=https%3A%2F%2Fzupimages.net%2Fup%2F26%2F16%2Fld8h.jpg 500 (Internal Server Error) After investigation, I found two root causes: 1. Direct image URLs being sent to the proxy The custom JavaScript responsible for detecting links and sending them to the proxy was using the following regex to exclude direct image links: var fileExtensionPattern = /\.(png|jpeg|gif|pdf|docx?|xlsx?|pptx?|zip|rar|svg)$/i; Note that .jpg and .webp were missing from the pattern. As a result, links ending in .jpg were not recognized as direct image URLs and were forwarded to the OGProxy, which then tried to scrape them as web pages using open-graph-scraper — causing a 500 error. The fix was to add the missing extensions: var fileExtensionPattern = /\.(jpg|png|jpeg|gif|pdf|docx?|xlsx?|pptx?|zip|rar|svg|webp)$/i; 2. The proxy not following HTTP redirects Some image hosting services (e.g. zupimages.net) return a 301 redirect from the bare domain to www. When curl follows the redirect manually the image loads fine: curl -IL https://zupimages.net/up/26/16/ld8h.jpg HTTP/2 301 → https://www.zupimages.net/up/26/16/ld8h.jpg HTTP/2 200 However, the proxy’s axios.get() call does not handle this gracefully when open-graph-scraper is involved, resulting in a 500 error being returned to the client. My questions are: Is there a known best practice for handling redirect chains in open-graph-scraper? Would passing maxRedirects or followRedirect options explicitly to axios or ogs fix this reliably? Is there a cleaner way to pre-filter direct image/file URLs before they reach the proxy, ideally at the NodeBB plugin level rather than in custom JS? Thanks in advance.
  • Clustering for NodeBB enabled

    Announcements cluster
    22
    1
    16 Votes
    22 Posts
    4k Views
    @Madchatthew True. I think this is the reason as to why most Open Source projects are abandoned because they are not sustainable in the long-term.
  • Do you need a degree to succeed?

    Moved Blog degree success experience
    5
    1
    5 Votes
    5 Posts
    1k Views
    @qwinter very well put. Great points and I can certainly align with these. I personally don’t see no university as a barrier to progression. My old boss said that he’d take preference with anyone who had a degree because of their “ability to think logically” (I kid you not). I said “well, you hired me and I don’t have a degree…”. He paused for a moment realising that he’d literally dug himself a hole and fell in it. He then said “ah yes, but you’re an exception”. “Exception” or not - it’s still a bigoted reasoning mechanism, and elitist to put it mildly. Class distinction springs to mind here.
  • Testing out Webdock.io

    Moved Announcements webdock recovery speed
    2
    5
    5 Votes
    2 Posts
    1k Views
    Just coming back to this thread for review (as I often do), and it looks like Webdock have increased their available offerings - some are extremely powerful, yet very competitive from the pricing perspective. [image: 1692559685163-7cf9a928-ac21-44fe-99c6-90439030d631-image.png] 10 CPU cores, plus 20Gb RAM? Well worth a look (and the asking price) - there’s also a fixed IP which is hugely beneficial. Clearly, this is well beyond what most people will want to spend - it’s more of an example (but interestingly, Sudonix runs on something not too different from the above). However, not all that glitters is gold - just have a walk through the benchmark report I found below and you’ll see a huge difference between Heztner and Webdock https://www.vpsbenchmarks.com/compare/hetzner_vs_webdock That being said, the amount of HTTP requests that Webdock handles in relation to Hetzner is superior - @DownPW you might want to have a look at this - there’s a free 24 hour trial… [image: 1692560710486-5203639b-2f62-47e6-b87b-37580ce5deae-image.png]
  • 1 Votes
    3 Posts
    1k Views
    @downpw I’m inclined to agree with the self-development - and the point you make about the lack of certification being seen as a barrier to enter professions. @downpw said in Experience vs Certification - who wins?: Because even without experience, if you have an iron will, you can be better than someone who has certification. With more work and effort, certainly, but does it matter? This is an interesting statement because I knew of several people years ago whom obtained MCSE accreditation, but had never sat in front of a real-life scenario before where they had to fix something - hence, the term, “Paper MCSE”. Essentially, this involves reading a book, then going to do the exam. Some people have that capability where they can retain a bulk load of information, and use that to easily pass an exam - yet have no practical experience. To me, it’s easy to identify these people. They can give you a text-book answer, but have no ability to prove it physically.
  • Fancybox now used for image handling

    Announcements fancybox
    16
    6 Votes
    16 Posts
    4k Views
    And it seems to be less conflicting!