Despite being live since 1997, OSAlert has had fairly few redesigns in the grand scheme of things. If my memory serves me correctly, we’ve had a grand total of 6 designs, and we’re currently on version 6, introduced about 5 years ago because of unpleasant reasons. It’s now 2024, and for a variety of reasons, we’re looking to work towards version 7 of our almost 30 year old website, and we need help.
I have a very clear idea of what I want OSAlert 7 to be like – including mockups. The general goals are making the site visually simpler, reducing our dependency on WordPress extensions, and reducing the complexity of our theme and website elements to make it a bit easier for someone like me to change small things without breaking anything. Oh and a dark mode that works. Note that we’re not looking to change backends or anything like that – WordPress will stay.
If you have the WordPress, design, and developer skills to make something like this a reality, and in the process shape the visual identity of one of the oldest continuously running technology news websites in the world, send me an email.
Have you considered converting the site to static HTML + Javascript with a static site generator like Hugo? It’s really liberating to not be tied to a PHP + MySQL backend with all of the cruft and upstream bugs and limitations and security vulnerabilities and upgrade hassles that come with server-side content management systems like WordPress. Then for the comments section you could either use a self-hosted Docker image (super easy administration and upgrades) that is specifically built for comment sections on static HTML websites, or alternatively you could even make it pull comments from a Matrix room or a Mastodon thread.
– Hosted and self-hosted comment solutions: getshifter.io/static-site-comments
– Use Mastodon as a comment backend:
* jszym.com/blog/mastodon_blog_comments
* jefftk.com/p/mastodon-replies-as-comments
* andreas.scherbaum.la/post/2024-05-23_client-side-comments-with-mastodon-on-a-static-hugo-website
– Use Matrix room as comment backend: cactus.chat
I think the whole point is to keep it simple on the back end while improving the front end design, hence the following from Thom:
I agree with Thom on this too; when the focus is the content, the back end needs to be as simple and maintenance free as possible, and WordPress is a good option for that. If they were to move to something like you’re suggesting, they would spend all their time maintaining the back end and we would never see any actual news. Despite its heritage, the site simply isn’t big enough to support a full time developer to maintain all of that. Regardless of what you said, WordPress isn’t a nightmare to maintain and there’s a reason it’s popular with creators who want to just focus on their content and not worry about hiring a team of devs to hold a house of cards together using the latest “here today gone tomorrow” web framework. If you’ve properly configured your WordPress server from the start, you don’t have upgrade issues, and if you stay on top of upgrades, you don’t deal with any of the other crap. I speak from experience having set up and maintained several WordPress sites for clients over the past two decades.
Yep, those are important considerations. But that’s exactly why I suggested what I did. There’s nothing simple or low maintenance about a server-side web app as complex as WordPress, and proof of that is all of the major hassles that we’ve seen here with things like comment editing plugins and even login problems that completely prevented posting (which was my case for a long time). The static site generators that I suggest are increasingly being preferred by content creators that want to focus on content and not technical baggage, just like you mentioned. Basically you write an article in Markdown format (which is ideal for structuring the article and focusing on the content instead of hassling with a janky WYSIWYG editor), and then run a single command on your local machine to automatically convert the article into proper HTML and publish it. Most static site generators have tons of free templates already developed and ready to use out-of-the-box. There is nothing at all to “maintain” if a managed hosting service is used, because it only needs a simple web server like Apache or Nginx to serve the static HTML files. You can even use something like Github Pages for free so as to not even maintain a web server. Interactive server-side processing is required for the user comments section, but again the suggestions I posted can use hosted services like Mastodon or Matrix that again require zero maintenance or further development.
That statement alone tells me you have no idea what you’re talking about. Mastodon, Matrix, Mattermost, any of the self-hosted discussion solutions like that require constant maintenance and updates as well. They are all in active development and require the same level of updates and maintenance as WordPress itself. Again, I speak from experience as I host a Mattermost instance at my workplace so we don’t suffer the limitations of Slack or Teams. Being a sysadmin is *never* a “set it and forget it” job. If it was, I’d be unemployed.
Your opinion and preferences are valid, but accusing others of ignorance is not a good look. I’ve also been administrating this stuff for decades. Read the links I posted, most of those options for comments do not require self-hosting.
@rahim123,
Fair enough, I don’t know you so who am I to say you do or don’t know your stuff. I only have what you have written here to go by, and I find it lacking in basic sysadmin common sense, but that’s just my opinion as you put it. *Someone somewhere* is maintaining those services, just because it’s not self-hosted doesn’t mean someone out there isn’t keeping it up to date. You’re just passing the buck on to someone you don’t know in the name of “simplicity”.
Saying a software package requires “zero maintenance or further development” as you put it is, quite simply, incorrect. That is a fact. By telling Thom and the OSAlert owners they should switch from what they know they can rely on to something hosted somewhere else by someone who isn’t maintaining or updating it, not only is a recipe for disaster, it makes it seem like you are contradicting your entire point.
I really don’t mean to be harsh, but I just don’t understand how you think such a convoluted, error prone, spaghetti mess of other people’s projects thrown together on a couple of Docker containers and some random person’s test-it-in-prod overprovisioned VPS Matrix server that could disappear tomorrow is at all a solution for a major news website.
Static HTML files are not vulnerable to exploits. A static site generator generates a complete website out of HTML + Javascript files for all of the articles, which gets uploaded to a webserver. So the only maintenance required would be routine maintenance to the underlying webserver and operating system, which can also be delegated to the hosting company. So that takes care of the entire website, minus the comments sections.
You are again correct, but the problem is that I didn’t say that regarding a software package. I specifically mentioned a hosted service. Here’s my full quote:
I interpret your position as rather conflicted, because on one side you suggest that the current method of self-hosting OSAlert on WordPress is superior, while at the same time viewing me as a nut job for suggesting self-hosting some components of the site’s functionality.
That would indeed be a disaster, and I never suggested that.
At some point in the stack you are inevitably passing the buck on to somebody else’s infrastructure. Specifically where that jumping off point occurs is a unique decision for each website, and if needed almost all of the open source projects offer their product as a professionally hosted service that requires little or no user maintenance. Of course even WordPress offers such a service, but the fact that it is a single project with all of the functionality does not make it any less of a “spaghetti mess” internally, and the proof is in the pudding that a WordPress instance is also not easy to properly administer and requires constant maintenance against a constantly moving target of upstream changes.
@rahim123,
I think at this point it’s clear we each misunderstood the other, because you are saying I said things I didn’t (I NEVER called you a nutjob and never would), and I must have misinterpreted your original comment calling for a mixture of static self hosted HTML and a hodgepodge of comment solutions based on other people’s chat servers. Either way, all of our mutual dick waving doesn’t mean anything to anyone since Thom has already said they are staying with WordPress, and my point all along was that that is the sensible thing to do.
I’m sorry, it wasn’t my intention to get into a flame war, and let me say again I don’t think you’re insane or any other ad hominem crap like that. I simply responded to what you said at face value, because in my opinion (there’s that word again!) your suggestions were flawed. Telling you I think you are wrong and questioning your methodology is not me calling you names. I could have been more tactful but as usual I ate my own foot when it came to giving out my opinion.
Morgan,
I’ve see a lot of websites contort themselves around wordpress when IMHO a custom solution would serve them better. I especially hate seeing data getting manipulated into wordpress tables that do a poor job of representing the data… this is very popular with wordpless extensions and I wish we’d stop doing that because the result is SQL queries that are both slow and complex to use. SQL is such an awesome tool, but let’s use it the right way! Haha. A properly designed database is so nice to use and can improve performance too….but then we come to your next point…
Ah yes. Ahem… that. Cost is the reason wordpress keeps replacing custom web developers. Not because the results are better, but clearly using wordpress is cheaper than hiring someone to build out/maintain a purpose built website. While I’ve gotten some leads to customize wordpress websites, their budgets tend not to allow for much innovation even if the ideas are good.
It depends on what you do with wordpress. Using wordpress out of the box is easy enough, but it’s not all cupcakes if you need to extend or customize it. WordPress can be limiting/insufficient for certain businesses/other requirements. Clients often say “I want ‘this’ changed”, but it can be really tricky when wordpress doesn’t have an option for it Maybe you can hunt down some 3rd party extensions, sometimes you have to write your own code. Creating plugins through the callback system can be tedious (and expensive) to implement. Typically these callbacks perform worse than a more strait forward php website would.
The trouble is wordpress updates can break local extensions/mods, sometimes it’s been my own, other times it’s extensions the client paid for that are broken, If the client is lucky, they can buy a new version of the plugin that’s compatible with their new version of wordpress. But if it’s discontinued you may have to start over with another plugin provider, which can make for a very painful wordpress upgrade process. Incidentally I suspect these kinds of wordpress issues were behind osnews’ problems. This is probably why Thom expressed the point of “reducing our dependency on WordPress extensions”.
I am playing a bit of devil’s advocate here. Custom websites are dead and I know it. A competent developer can do better than wordpress, but I don’t expect osnews to fight the tides. Not only is it a question of resources, but the interest isn’t there either.. I am positive they are thinking “the less self support we need, the better”.
That said though, if osnews have any special feature requests, we’d probably have enough talent here to make it happen. Don’t shy away from asking your users
Make dark mode actually working.
Thank Cthulhu and the OSAlert team for the edit button being back! Sorry zdzichu, didn’t mean to reply to you.
Dark Mode, please!
I’ve been following Ars (back when it was still Ars) and OSAlert since their very early days. One thing I can say about Ars (despite not being Ars any longer) is that I can still revert to the original design: white on black, time-based listing of articles – and the full Dark Mode is supported in the comments as well.
IIRC my early HTML5 and CSS stuff this should be relatively easy to implement here (I’m thinking a switch from a white CSS to a dark CSS, but nowadays I might very well be wrong). The layout here on OSN basically follows the original time-based Ars layout, so there shouldn’t be any problems there.
I *swear*, if I have time I’ll follow up on this post with some sort of mock-up.
You can do it Thom!
Can I put it out there that I think it would aiming for AAA WCAG would be a really positive thing. To often people with accessibility needs are forgotten in the drive for more functionality.
Using only HTML standards would also be nice!
NFR FTW!
I signed up so I could add to this thread after I suggested something similar to Thom over email.
WordPress is smart enough to allow any front-end, including static sites. I have great experience writing Gatsby (React) front-ends for WordPress. The advantages are that we get to keep the back end without any modifications, we could get rid of all of the plug-ins, and the front end will be blazingly fast, with practically no upkeep, and very cheap (free?) to host.
If Thom is interested in exploring further, I will gladly set up a demo template of the site to showcase these capabilities.
Not to disparage your work elsewhere, but $DEITY preserve us from a Gatsby or even plain React frontend for a site as simple as this one. The idea that such a website is ‘blazingly fast, with practically no upkeep’ is laughable, considering the web development community has finally realized how slow React is and remembering how much fallout there has been from React updates, to say nothing of the JavaScript ecosystem’s disregard for stability. I’m personally a big fan of SSGs—Eleventy being the best of the lot in my opinion—but even if the article didn’t explicitly say ‘WordPress will stay’, Gatsby and React are among the last options I’d reach for.
This thread is microcosm of open source software development.
PO – “I want to implement this with this tech”
Dev 1 – “but you Could use this instead”
Dev 2 – “Dev 1 is wrong, but you Could use this instead”
Dev 3 – “Dev 1 and 2 are wrong, write it in rust”
Adurbe,
Haha, true.
While I think spiffy things could be done on a custom site, honestly I don’t think the rationale for choosing wordpress in the first place have changed. Osnews isn’t in the business of writing code and doesn’t want to be.
Personally I don’t really care much about themes, functionality matters a lot more to me. But I know people have very different priorities when it comes to website design.
I would argue that there is a general agreement that WordPress is bad and should be replaced where easy to do so (front end) and that static site generators are probably the best solution.
Which SSG exactly isn’t really important in my mind, in the end of the day the suggestion is a static front end that will be fast and robust instead of the WP one.
…I wonder if there’s an SSG written in rust…