Search results for:

As attachments, there are none... the forum does not include a player.

But the forum will embed players from other platforms, so if you paste a link to a YouTube video - i.e. you uploaded it there - then the forum software will convert the link into an embed of the YouTube player with that video showing.

i.e. merely pasting this into a comment:

https://www.youtube.com/watch?v=epEGN-7GsPs

Will result in this rendering:

https://www.youtube.com/watch?v=epEGN-7GsPs

Glad you have a workaround.

There seems to be no inherent reason why the forum isn't displaying it correctly... the forum is UTF-8 end-to-end and manually typing it is just fine.

It makes me wonder whether the process of copying and pasting from the console into the comment has done something, whether the console itself isn't displaying as full UTF-8 and so it doesn't copy as such.

As it works (seen above in my comments) I don't believe this to be a bug with the forum, but I am curious as to where it creeps in.

Oh, it's in a source code quoted part... interesting. No, not seen that before.

EUR = €
4,98€

I'm flummoxed.

€

Dealt with the other spammer... more safely too.

Messages should now be restored.

Dealing with the other spate of spam now.

Oh... and I could construct knowledge that a huddle existed from the comments... i.e. who commented = who has permission, and the comments know which huddle they belonged to.

The missing thing is I don't know the title of the huddles.

Oh well, this backup can't take forever.

Will they be restored?

Yes!

So the messages are still in the database, but I screwed up a query when I nuked the spam.

For reference a "huddle" is kinda like a direct message conversation.

The data structure is "huddle exists and contains comments, huddle profiles maps huddles to those who can view them and contribute further messages"... the profiles exist, the comments exist, but when I nuked the spam I figured I had left orphaned huddles so best delete them. I wrote a complex piece of SQL to "find huddles which have only deleted comments, remove the profiles and delete the huddle"... but the bit that is "have only deleted comments" was wrong, and the query went too far and has deleted the huddles for all on this site.

So... how to repair? Well... just get the valid huddles and huddle profiles bits back from a backup.

Where are my backups? Tarsnap.

I've been pulling a backup all day, but I back up the whole database and so far it's at 14GB downloaded and decrypted, and then I need to pull it somewhere and restore the two tables, find the rows to be copied back and then put them back.

And lesson learned? Don't delete spam manually even when I don't have access to test environments... ban first (safe) and delete later with a proven and tested script.

Not that I know of.

I typically report them to stopforumspam and this forum uses that database as well.

I've banned the user... but as it's a human, with a Gmail account, and doing this manually... it's not proving to be a very effective ban.

I'll be looking at what I can do later this evening to improve this, and to make the bans more effective and whether I can put in logic like "if you haven't posted publicly you can't do more than n private conversations"

It's worth saying, auth is driven by your email address... nothing more.

If you have a Google account and were signed up with the same email... then press that button, it works great.

If you have a Github account and the email on that is the same as the email you used with Persona... that also works great.

The email bit is the "this will always work, given the same email". It sends an email code to your email address immediately, that contains a TOTP 6-digit token that you can type in to gain access.

Email is how we identify a user... as long as that matches you're all good.

If any person finds that they have logged in and got a new account and a "user76588" style name, then send me a PM and I can take the email you have now logged in with and assign it to the old account (you need to tell me what account that was, I'll check things before proceeding). I'll basically make whatever your preferred auth method is, and will make it work for you.

I can fix that.

Should've raised a github issue for me to see. But as I've found this I'll add it for you and will fix it for the next release that is coming in a week or two (adds subforums, new permissions management, fixes bugs, improves search, etc).

Track this bug here: https://github.com/microcosm-cc/microweb/issues/286

Saturday's are for fixing bugs ;)

We're using https://github.com/russross/blackfriday for markdown processing and it has a range of flags to enable/disable various features.

We've now enabled Github Flavored Markdown for things like code blocks.

And I've since disabled SmartyPants which attempts to do horrible "smart quotes" and which was turning single quotes into question marks.

We're very receptive to any and all feedback, and are monitoring all of the sites we're running for issues. So feel free to let us know via the https://meta.microcosm.app site of any issues you find.

We'd rather not pollute Gordon's space with forum questions, and keep this space focused on Espruino stuff. Especially as both @motter and I were in on the Kickstarter and are looking forward to nerding out and seeing what we can do with those.

I learned it for the web, mostly when mucking around with forms.

I had an idea back in 1999 that I could make a search engine that didn't require a search engine. It was for football videos, and the form had a load of SELECT lists that I wanted to filter each other, such that you searched by choosing things: "This team, in that season, against this other team, at that venue, show me the goals and fouls.".

I kinda knew it was possible, but at the time that meant writing my own cross-browser library to figure out the differences between Netscape Navigator and Internet Explorer.

Then I went mad on this new-found knowledge and decided that what the world really needed were IFRAMEs. Except they didn't really exist at the time, and so I used a frameset to hide a 1 pixel high frame in the bottom of the window.

Then I made the JavaScript in the main window detect changes in a SELECT list, and load a pre-defined JavaScript file in the hidden frame. Once that frame loaded it would reference a function in the main frame as a kind of callback mechanism, and populate the next SELECT list.

I sat down and did the math, and if I just splatted 310,000 JavaScript files onto the file system to cope with every permutation of SELECT list, then I would've created a search engine that used filtered lists to search and was driven totally by static files.

It took a few days to write, and more than a little smashing of my head against the wall as I quickly discovered the quirks between Netscape and Internet Explorer, but to my total shock it worked.

Not only did it work, but it worked well. When the site was eventually launched across 300 football club sites, this one part of the site was heavily promoted and had hundreds of thousands of concurrent users in the first weekend. Because it was all just static files, nothing broke at all.

I liked JavaScript after that, probably went a bit overboard with it for a while, now I use it sparingly but quite enjoy it when I do use it.