Wedge

Talking with Pandos reminded me that I needed to open a topic about this.

So, a few weeks ago, I wrote a quick, rough implementation of relative dates for Wedge. They behave like in Facebook or other 'modern' software, or, closer to us, ElkArte. I wanted pages to validate, and still use valid semantic tags, so I adopted the time tag, and used DATE_W3C as the datetime format, it's longer but more 'solid', and takes timezones into account.

At that point, I had: <time datetime="w3c valid date">Absolute date</time>
Okay.

I used JS to transform absolute dates to relative dates, and it's pretty neat and tidy. I'm happy with it.
But JS has this tiny little problem: it doesn't execute immediately (whatever you may think), and as such, the browser first shows the page with absolute dates, then executes the JS, and replaces dates with relative dates.
The problem is that I didn't realize this because I usually do page reloads, and Chrome is smart enough to execute the JS ASAP and thus I never see the absolute dates. However, if I simply move to a new page, I can see the absolute dates for a split second. I don't like that.

The solution would be to then use relative dates right in the HTML, but what if I want to know the exact date a message was posted at..? I'll have to keep it in a title tag (when right now, the title tag is generated on the fly from the contents of the time tag. Smart enough.) So, that makes this kind of line: <time datetime="w3c valid date" title="Absolute date">Relative date</time>

And, believe me, I think it's wayyyy overkill just for providing an accessible way of getting a relative date. I was already unsure about committing my HTML/JS work with the first version, but this one... Well, it's just not cool.
Not only that, but what if the user likes absolute dates better? Oh, well, I guess I can then simply change the HTML I'm printing out, I guess... I can wrap it up inside a small helper function, provide it a timestamp and then expect Wedge to do all the work for me. Okay, I'm okay with that. I can do that. The only thing I'm not looking forward to, is adding the UI for it, even if it's simple, because I'm rather remove user options than add them, but well, I think it's something that polarizes users more than "number of topics per page", I'd say.

But the extra length of time tags, I'm not sure it's worth it... :-/

So, two questions... Do you like absolute or relative dates better, and do you think it's all worth implementing (considering if I do, I'll add a setting to let you choose your preference)?
Also, when it comes to relative dates, how far away (ago?) do you think it's worth reverting back to absolute dates? I'm not exactly the kind of guy who enjoys seeing a "6 years ago" date. I prefer to see "May 2007", but maybe that's just me. I also don't like seeing "3 months ago" either, but I'm fine with "5 days ago", so... I guess it's a fair question.

PS: oh, another one... I also think it'd be really cool to see "5 days ago at 8:54" rather than just "5 days ago". Oddly, I've never seen this kind of implementation, anywhere...

I'm doing this topic in the public area, in the hopes that:
(1) my issues may be resolved by more people,
(2) anyone can learn, too, and later be able to contribute to the repo when it's made public.

So... Wedge is now a git repo (yaaaaay...), and leaves SVN behind. (I'm not completely finished with it, as I plan to restore revision numbers through either tags or notes, whatever is best, in the future, when I have a bit more time to devote to that.)

Git is a complicated beast to master, and I'm only getting started.

Notably, what needs to be 'understood' is how communicating between repos is made. Since it's not stored on a central server, no repo is the 'right' repo, so I consider the right repo to be my local repo. For all intent though, it's exactly like the BitBucket private repo I have right now, since my 'origin' ref is set to that one.

There are 4 matters that I wish to solve before I can continue. (Maybe 5, but I can't remember the last one, it'll come back to me.)

1/ When doing a pull from my remote origin, TortoiseGit fails because some files I've edited locally were changed in the meantime on the remote repo. It told me to stash my uncommitted changes, but I can't find a stash command in TortoiseGit. So, I fired up SmartGit, and tried to pull from there. It offered me a choice between merging and rebasing. I chose merging as it's the default Git behavior, but they recommend rebasing. However, I want to keep branches, to make it easier to find areas that were pulled from another branch obviously. Apparently, it was the right choice, because my logs are identical between HEAD and origin/master, so it looks like it copied the history straight from the remote origin, right..? Still, I'm lost, I'm not sure what's best for communication between my repos.

2/ In my current situation, I'm pushing my own changes to the remote origin directly, when I could be doing a repo on my account, and then push it to the team account's repo. However, I don't know how to do these pushes manually or automatically, especially on a remote server, so... A bit stuck. FWIW, I don't know how the current collaborators (Pandos and TE) can merge their own branches with the updates on the official repo. Honestly, it eludes me. I need some frigging documentation on that, I guess.

3/ Bitbucket sucks, somehow. They won't allow me to add a fake e-mail address to my account, so I can't associate nao@wedge with my own account. Sucks. Big time. However, since it will end up at github eventually, it won't be a problem, soon. However, my biggest problem is that Bitbucket uses my 'actual' e-mail address for web-based commits (e.g. pull requests), thus I currently have, in the online (private!) repo, my e-mail address in the clear. I don't want that. Github provides no-spam addresses, so I'm okay with that, but BB doesn't have anything. I'm forced to commit with that e-mail address, no way to conceal it.
I tried 'amending' my latest commit, but it didn't work. I don't know if it's because the latest commit is not technically from my git account that I can't seem to actually amend it, or if it's just how it works (doesn't work). I did 'git commit --amend --reset-author' in a bash window, and it simply launched a Vim page with the commit data. I didn't know what to do, so I just closed the window. TortoiseGit didn't offer to amend. SmartGit offered to do it, but I can't seem to make it work. Finally, Git Gui did the amend as I asked, but it didn't change the author at all (couldn't find an option to --reset-author, which probably didn't help.)
Anyone knows how I can fix that particular commit....?

:edit: Actually-- I just restarted my TortoiseGit log (which I'd only 'refreshed'), and it marked the commit as using my dummy e-mail address. Yoohoo! Well, it's still not cool that I'm going to have to amend all of my future pull request merges, but... Well, it's an extra incentive to hurry up and go public, ah ah...!

4/ Right now I have 36 uncommitted files, with three thematic changesets: privacy rewrite, contact list stuff, and miscellaneous one-bits. Is there a way to split these into several branches, so that I can safely work on them separately, and then commit as needed? Or do I absolutely need to create a branch, then start my stuff? I'm guessing I could always 'stash' the changes, then create a branch, then unstash..? (But then I need to selectively stash for each file, which I don't think is possible, ah ah...)

Okay, that's all for now...

I'm a bit annoyed that I've been spending the better part of the last 3 days working on the git innards, instead of actually committing my private rewrites, but... Whatever. I have to do it. It's the only way Wedge can keep up, at this point. It's so easy to do pull requests, I want to enable anyone to do them. I'm just hoping I can live up to it, see..? What's a fantastic git repo, if the maintainer isn't able to do the basic stuff...? :P

So... It's been over a year since the last blog post, and a few weeks since we celebrated the 3rd anniversary of the creation of this little project called Wedge, that took the life of two promising web developers and turned them into lifeless zombies with a crave for blood. Oh, maybe I meant vampires.

For those of you who didn't follow the latest developments on the forum, I'm sad to announce that one of the zombies is now officially resting in peace, carrying his own head, basking in the glory of his past achievements. Said zombie we called Arantor, and in pure retirony style, he decided to call it quits a couple of days before the anniversary. And I'd planned to show up at his door on that day and give him cookies! Well, no cookies then. Blame that good old UK craze about cookies.

And here we are. It's lonely at the top, but at least it means I get to make some hard decisions that we always postponed because we weren't sure about them. Same for Pete on his side, I'm guessing, albeit with other projects. I'm hoping we'll still see him pop up from time to time, and share his disbelief about how badly I handled the project. I always appreciate constructive criticism, after all.

Now, here's the point I wanted to address. There you go, Pete's state of mind up to this point was that we had no plans to release Wedge in the near future, that we didn't even have a tentative release date, and that it was even unlikely to ever happen. I understand his position, I'm not going to put the blame on anything; fact is, yes we slowly drifted apart as collaborators in the last couple of years, mostly because of real life issues for each of us, and because the weight of the project started to get very heavy year after year. The longer you wait until release, the more pressure you put on yourself to make it worthy of the wait.

And this is my first decision.

I'm putting Wedge on the road it should never have left: the road to release. Not the road to completion, which is a different thing altogether. With such a huge project, one man alone just can't complete the project. At most, he can do 80% of it, but he has to rely on other people to do the rest, simply because there are always some tasks where you know that other people will do a better job; that's what kept Pete and I together for so long, and I was hoping it would remain that way, because we really were complementary on most points. Working alone is not easy. Some days, you just want to say "screw Wedge!", and watch a movie or a TV show for hours. And you know what? I'll do just that. I've seen tons of silly and better movies, and I'm still going back to Wedge every day, because if there's something I'll never, ever give up before it's out for the public to enjoy, it's Wedge. Even if it doesn't bring me a dime, I need to finish it.

So, what's the tentative release date? Nope, sorry... There still isn't any. And before you ask why -- ask yourself whether you prefer to be given a wrong release date, or be told exactly what remains to be done to have a releasable product. To the former, I'll just say this: I'm hoping to release a public beta before the end of the year, and a final 1.0 in early 2014. Do what you want with that silly date, and if you're in too much of a hurry, and you consider yourself to be a hardened, hardboiled, hardcore, hard-wired forum administrator with a taste for trouble and a love for HTML/JS/CSS/PHP/FTP tweaks, you can always contact me with a request for private beta access and a copy of the SMF importer code.

To the latter, I'll be more thorough. First, I'll try to list a few of the things I wanted to do since day one, and will probably not do in the first release... Aeva Media is a monster to handle, it used to be my full-time (unpaid) job. I wanted to redo so many parts of it. Most importantly, I wanted to replacement attachments with media items; I'm still planning to do it, but if it's not done when everything else is ready, I'll still release. Also, Noisen.com features... I wrote many features for my personal website around 2007-2008, and was adamant I'd include them. Pete persuaded me otherwise, for very good reasons mainly, but now that he's gone, I can decide, if I want, to fail in a spectacular fasion. Go get the pop-corn. Again: if it's not ready when the rest is, I'll still release.

And now, for the features that I can't see myself release without.

- Privacy (estimated amount of work: a few weeks) : there are already some privacy features in; thoughts have privacy. Topics have some sort of privacy. I'm hoping I can do board privacy, à la Noisen, and profile privacy, à la Facebook. And fix an annoying bug in the 'Contacts' option: if you post a thought for your contact list, and someone replies to it, and you're not in their contact list, then you won't be able to view their reply. To be honest, I've thought of several solutions for that, but it's hardly anything solid, or at least efficient.

- Contact lists (a few weeks): so far, Wedge has been using SMF buddy lists, while Noisen has been using a customized version of Ultimate Profile (a nice little SMF mod from back in the day), where I allowed you to 'hide' a contact from view, and nice little things like that. Contact lists are my biggest project right now, I've already started work on them, but still need to write the UI, and some other things, to make them usable. The idea is that you can categorize your friends into multiple lists, choose the list name, and even determine the visibility of said list to profile viewers; you should be able to add people to multiple lists, and target a thought/topic privacy to a specific list. (Not multiple lists -- it's already complicated as it is, UI-wise.) My biggest issue right now, is handling the aftermath of adding someone to a list. Should they be notified of it? Should a list have a setting to determine if notifications should be issued? Et cetera, et cetera...

- Skins (a few weeks?): there are two things here.

(1) I'm not sure whether I should enforce sub-skins to be put inside a parent skin's folder (which I always found to be a natural thing to do), or simply require them to specify the name of their parent (author:skin_name) in a skin option. The latter option would allow skins to have folders for images, templates, etc. without the need for a different folder naming scheme, such as adding an underscore (_) before the name of any folder that isn't a sub-skin.

(2) As I discussed recently in the Friends board, I'm very, very tempted to remove themes from Wedge entirely. Just like Pete removed SMF packages when he finished his implementation of plugins, I'm in a similar situation where my skin system can do 99% of what templates can do, and it's just tempting to say, "themers have to re-learn mostly everything anyway, so I might as well get rid of that relic." Recently, I implemented one of the last things missing from skin features -- the ability to override/before/after a template function from within a skin.xml file. It was already possible to override a function through the use of a Custom.template.php file inside the skin, but it's not a rock solid solution because of the possibility of conflicting function names, and I only added this as a way for themers to 'test' their replacement code in a comfortable PHP editing environment, before they simply move their functions to an XML editing environment, without the proper syntax highlighting.

- AeMe comments (a week?): I'm planning to remove them entirely, and replace them with regular board topics. The idea is to create a generic board (hidden or not), where topics are created on the fly when posting a comment to an album or a media item. The advantage is that you get to comment on whole albums as well, and that you can search inside media item comments. The drawback is that I'm not planning to write an import tool for existing comments. They'll either have to go, or Thorsten (@TE) will have to account for them in his fantastic OpenImporter tool. Oh yes, and another drawback-- I'll have to handle attempts to access a media comment topic directly, by redirecting to said media item (or album). I don't know how I can do it efficiently, though.

- Others (weeks? months?): There are, I'm afraid, many other things that I'll attempt to fix before I go public, but none of these should be as big as the aforementioned items. When added up, though, they will probably take more time than them to complete, and there's always the tiny issue that my to-do list always seems to grow twice faster than I can shorten it. I'll have to pick a time when I figure that, well, it's okay Nao, it's okay... You can release.

With great power, comes gr... Oh no, not that Spider-Man crap again. But still, Wedge is an important project, one that could bring a nice breath of fresh air to the forum world, and even to the CMS world. As in, "community management system", not "content". I'm devoted to making Wedge a great, fun way to communicate with people, be they your friends or ephemeral guests, and help them feel at home and find the answers they seek.

And if it works, then maybe, in some way, it wiil also be the answer I've been looking for.
Get to work, Nao... Now!

Story of my life...


As a result, the wedge.org homepage shows much smaller characters to me in Chrome for Android, and it's like I have to restart from scratch... Obviously, I don't want to play the 'guess' game, and I'd like to ask you guys to... post a screenshot of your wedge.org page being rendered in your mobile device.

- You can either choose the wedge.org homepage, or this very post. It would make more sense to choose this post, as you can take a screenshot right now... ;)
- You can use Chrome for Android, or the stock browser on your device. Not anything else, please. You can do both, if you want.
- If you have the same results as someone else with the same setup who posted a screenshot before you, please don't post your screenshot, just confirm or like their post.
- On Android, the screenshot shortcut is usually Power + Volume Down. On some devices, you need to swipe your entire hand (the pinky should pretty much touch the screen, and the hand should be perpendicular to the screen), from left to right, to generate a screenshot.
- On iOS, it's Power + Home.
- Just attach the screenshot in your post.
- Indicate: your OS and OS version, your browser name and version, and any changes you may have applied to your setup (e.g. whether it's a custom ROM, or you did something to change the font face, or size, etc.)

This should, hopefully, help me determine what's the 'correct' size right now, and what I should be aiming for.

It would mean 140k, though... :^^;:

As mentioned in the thoughts (private ones[1]), someone has been having trouble logging into the forum.
I looked into it, reset their password, and it still didn't work... They got a brute force warning.

Now, here's the thing. I remembered that Pete had committed some SMF fixes for Wedge, regarding brute force issues, and I'm thinking, that's where the problem lies -- in this fix.
So, is it an SMF bug, or a Wedge bug..?

The SMF 2.1 rewrite:


The Wedge rewrite:


From what I understand: first one means, "if recorded flood timestamp is older than 20 seconds ago, reset it to 2, otherwise don't touch it.
Second one means the exact opposite... Meaning, once the flood record is older than 20 seconds, the user is forever locked out of the forum.
It took me an hour to finally find the culprit, and I'd like to know what exactly went into this rewrite..?

Also, the Elk version is cleaner, and much shorter...


Finally, a note about the code that tests the timestamp and resets it if it's too recent.. (?)

In SMF 2.1:


(Means "if time stamp is older than 10 seconds...")

in Wedge:


(Means "if time stamp is more recent than 10 seconds ago...")

I'm inclined to believe the Wedge version is the correct one here, but honestly, I have no fucking idea. I'm just not comfortable enough with the whole validatePasswordFlood function, to be clear. I understand what it does, but not the magic of it.
So, err... Need your opinion, guys..?

Hopefully Pete sees this topic through the magic of 3G, or something... :^^;:

Additional note...

Looks like I can login with my hardcoded flood value, but it doesn't get reset to nothing, so I'm still with the same flood value after that... (?!)

1.	I thought that using a public topic for this was okay, because I'm not going into specifics with the source code, and it's close enough to the SMF version anyway, which is freely available on github, so...

So, during this weekend, I found myself in an interesting position. My Windows 7 started behaving like it's 1995.
Basically, all of the fonts somehow forgot how to show Unicode characters. That is, I have plenty of Japanese programs and folders -- they show up as squares in the Start Menu and in my File Explorer. I was using PT Sans as my main font, so I changed it to my alternative preference, Tahoma, with the same effect. I tried many, many fonts, and it seems that only one font recognizes Japanese (and CJK in general) characters, and that is Segoe UI. Heck, even Arial and Times New Roman don't work, eh..?!

I've temporarily set my font to Segoe UI, but even though I love it for web design, I hate that font for daily use, I find it too... stylish, and not as well hinted for my monitor as PT Sans is, for instance.

I've done countless web searches on this issue, and it seems I'm not alone, but I really can't fix it, and was wondering if anyone had ever had the same issue..? Of course, not everyone needs Japanese support on their system, but whatever...

Maybe a reboot will fix it, but currently I'm still working on my NAS setup, so it's a no-go.

That weekend really, absolutely, pathetically SUCKED. -_-