I think they're broken. On #Windows anyway. As far as my logic understands, any running program should not know that they even exist. But #xampp can tell. Or at least, I think Windows is just feeding it directories as if it could tell. This is a little rough to explain.

Say we have a #PHP file on an #Apache server in a directory, but that directory is actually what Windows calls a junction (A type of symbolic link). If I 'require' another php file with a relative link like '../', where is the server going to look?

My first guess? The directory above the junction. It took me nearly an hour to determine that it was looking in the directory above the TARGET of the junction. Maybe that's the way that should work for some reason. But I do not know what that reason is. If Windows had any kind of proper sandboxing, it'd be a major security hole.

I do seem to be good at finding the #CornerCases.

I learned about a new #CSS attribute that's pretty cool: will-change.

/* Warns the browser transform will be updated a fair amount */
will-change: transform;

It basically tells the browser "hey, this attribute's probably going to change on you, so you better be ready." I don't know exactly how it works- if it pre-processes most likely updates, passes processing to the GPU, etc- but it fixed some problems with moving elements smoothly on mobile for me.

MDN actually says it's "intended to be used as a last resort" which is always a good sign for a new web standard IMO. It's good to keep our devices on their toes.

I'm betting we'll see it being abused to heck if it hasn't been already. It may just become the new !important.

#WebDevelopment

Chrome wants to let you cast any video or audio from a phone. Even if you don't want it to be castable.

Fortunately, the fix is pretty easy. Just use:

video.disableRemotePlayback = true;

This will prevent a cast icon from showing up on videos that you don't want your users to be able to cast, like background videos.

Why isn't it instead something like

video.remotePlayback = false;

though?

Probably not for a good reason. It may be that since unset booleans default to false in most programming languages, this was easier to implement and have videos default to being castable. But I feel like it's pretty sloppy.

That's the fix anyhow.

#WebDevelopment

Hey! If you're reading this, then you already noticed, but the #blog is now located at blog.bogedinbr.com and bogedinbr.com now has a bit of a front page for our development services. That page could use some work, it's a bit of a process. Of course I say "our" because #Josh and I are going in on this together. Should be fun!

Does the blog need to be a #subdomain? Probably not, but it's #cool right? Just another thing about #WebDevelopment that I understand even better now.

If you copy and paste the below into the dev console on most #websites:

var script=document.createElement('script');
script.src='https://joshpowlison.com/wut/havoc-button.js';
document.head.appendChild(script);

You'll get a nice button in the top-left corner of the screen that reads "Havoc".

I needed to do something silly as a break from my serious work, so I made this. It adjusts the CSS of elements on the site at random so they get all wonky. You can click the button multiple times, and it can even affects its own CSS.

Have fun!

Most of my #WebDevelopment work has been fixing broken #websites. And the first thing I do on any of these projects now is to start removing whatever I can.

Buggy #Wordpress site? I start deactivating plugins. If the website breaks further, I reactivate the plugin temporarily, but so far I've always been able to remove at least several plugins without it impacting the website at all. Usually the plugins that do impact site functionality can easily be replaced with some vanilla webcode.

Buggy non-Wordpress site? I start replacing #frameworks with vanilla webcode. It becomes easier to read, less convoluted, and it's easier to integrate the external code that really does matter (like Google Analytics code).

I call this process Subtraction. To me, it's an all-around win: the code is easier to read and faster to edit for me and future developers, and the website runs faster. Often Subtraction will solve the problems in and of themselves, or simplify the problems far enough that a solution is easy to implement (rather than trying to work around existing, conflicting webcode, which is often a problem).

Now, my first instinct on any project is to see what I can delete, and move forward from there. And I've always been glad that I took that step.

What up! I'm Josh.

I like carousels. But not on a website. It feels like the creators of the site couldn't decide what information was the most important, and so tried to make it *all* super-important.

This site expresses the idea pretty well.

I have the habit of scrolling past automatic carousels so they don't distract me while I'm web-surfing. Usually, the constant change just distracts and frustrates me.

Manual carousels? I'm usually not bored enough to bother.

Also, I'm going to try to break Brian's #website with foreign characters. Because that's what friends do.

οικω
amígo
こんにちは

#WebDevelopment

Have you ever wished you could have multiple folders on your system that contain the same files? And update in both places automatically? No? Oh, well then don't bother reading the rest of this.

For the cool people, symbolic links allow an operating system to treat files or folders as if they exist in multiple places. This is very useful for organization. In my case, I use them a lot for #WebDevelopment on my local server. I want to place all related project files in one place, but I need the actual web site to sit in the folder of the local server. Thanks to symbolic links, I can have both without manually updating the local site.

This works on a lot of operating systems, but for info on using them on #Windows, check out this HowToGeek article. I mostly use junctions to create folder links for entire websites, but the standard link type is also useful for single files. Just be careful, because a single file symbolic link looks EXACTLY the same as a shortcut in windows #explorer. But they are very different, as you can see if you check the properties. Also, the /H hard link seems to only make a copy, without auto updating the contents, so I don't ever use that.

You're welcome.

#tips #productivity

I found this interesting, so I'll share. There was a #bug in the code that decides how many pages of #blog posts there are. If there was an exact multiple of the max posts per page, there will be an extra, blank, page. Simply subtract one from the number of results:
$maxPage = intdiv($totalResults -1, $resultsPerPage);

Not complicated, but it took me a while to come up with this clean solution. No conditionals, it just works. For reference, this is the bit lower in the code for the older posts button:
if ($page <= $maxPage){ write the button }

I haven't been using #PDO and #MySQL for a very long time, so things are bound to pop up from time to time that confuse me a little. While developing the system for storing #hashtags in a table, I tried my hand at using a MySQL variable for the first time. In hindsight, it looks as though it wasn't necessary, but it's good practice anyway.

So, I wrote a statement that looked like this:
$stmt = $dbh->prepare('
INSERT INTO blog_posts (title, safe_title, body, created, author) VALUES (?, ?, ?, NOW(), ?);
SELECT LAST_INSERT_ID() INTO @post_id;
'.$hashtags['tag_insert'].';
');

$hashtags['tag_insert'] is another insert statement generated by the function that parses for tags in the post. It uses the @post_id variable. So, I'm not actually selecting anything in this statement, other than into the variable. But, trying to get the LAST_INSERT_ID() from this statement, either by adding an additional line to SELECT it, or by SELECTing it with another statement, does not work.

If you SELECT it with the same statement, fetch() doesn't return anything, but if you create a new statement, you get an error saying you "Cannot execute queries while other unbuffered queries are active" and that you should call fetchAll(). But there's nothing to fetch, so it doesn't work. So we need to deactivate that query some other way.

The solution is to use $stmt->closeCursor();. Whatever PDO thinks you might want to do with the statement, it just drops it. Now I can create another statement for the LAST_INSERT_ID(). And it still remembers the value of the variable @post_id, so I can just use that.

Here's a link to the Stackoverflow question that I read to solve my problem.

Weird. But I imagine it only seems that way because there's some insight into MySQL that I'm currently missing.