Wedge

It is, but I've come to really love it in the last couple of days.

Thanks, but I'll grab it myself shortly as part of a bunch of apps I want to get off iTunes. (Just that due to a screw-up at my bank, I'm not able to use my card that's on the account and when it's reactivated I have several things already to purchase)

Revision: 574
Author: arantor
Date: 15:56:48, 08 February 2011
Message:
! Tweak to licence wording for our protection (license.txt)
! Sync to SMF 10446 aka "Clusterfuck"
 ! (10404) Minor improvements to integration hooks. (Admin.php, Load.php, ManagePermissions.php, ManageSettings.php)
 ! (10405) [2.1] Added detection for Litespeed as part of the server config. (Load.php)
 ! (10408) Tabs within code tags were not parsed correctly for some browsers. (Subs.php) [Bug 4578]
 ! (10414) Validate odd URLs (QueryString, Subs.php) *See notes
 ! (10415) Oops. Remove completely incorrect protection against $cachedir removal. (upgrade.php)
 ! (10419) Update the change in 10414, so that password flood is validated regardless of the number of times before it. (Load.php)
 ! (10439) Save a query on posting (topic=161621) (Post2.php)
 ! (10441) [2.1] Don't log robot views on topics or attachments. (Display.php, Dlattach.php)
 ! (10442) sprintf call for support resources language string was overridden. (Admin.php)

Not implemented
* r10406 [2.1] Moves the admin helper functions out of ManageServer and into Admin. I'll be doing this in the rewrite (I'm not so bothered about syncing 2.1 changes back to us and there are potential licence issues anyway with just using 2.1 branch code). There is also a tweak to allow for $context['body_id'] to be defined and if defined, attach that id to the body tag. I left it off because I'm not really sure where this would actually be useful. There is also a lot of changes for the language editor from/to display sanitising but I have absolutely no idea why it's changed (since it's not referred to at all anywhere I can find)
* r10407 [2.1] parsesmileys() created a regex too complex - I'm sure this has consequences I don't like, while it has performance, it's also more naive in how it approaches things I think (it's discussed in the beta board, and Unknown was trying to talk groundup out of it more than once) and it's more == to === conversions to safely ignore.
* r10409-10413 Minor fixes (1.0 SSI, 1.0 patch file, 1.1 spacinazi, 1.1 SSI, 1.1 patch file)
* r10416 Changelog update
* r10417 The change in 10414 to sanitise $_REQUEST['start'], I already added the check that this rev adds (that $_REQUEST['start'] is defined before trying to (int) it)
* r10418 Changing the SSI security thing AGAIN (in line with what 10409-10413 did, basically to ensure that the kickguest only got called when appropriate, but we don't kick the guest to login anyway, we do what makes more sense in the first place: restrict the user's access in SSI so they can't see anything they're not supposed to and let the folks who use SSI figure out what they want to do with the user thereafter)
* r10420-10437, 10443-446 Updating 2.0 patch file, spacinazi and patch file, 10414's fix backported to 1.0, and for 1.1, file versions for both, 1.0 patch file, 1.1 patch file, 1.0 cleanup, 1.0 patch file, 1.1 file version typo, 1.1 spacinazi, CRLF issues, more CRLF issues, 1.1 patch file, 1.0 language file version, 1.1 SSI examples, 2.0 patch file, update to manual to match the trademark slogan of SMF, 1.1 language file version, 1.0 patch file, 1.1 patch file, 2.0 patch file
* r10438 [2.1] Caching members online stats is fine but I don't like the way it's done (should really be done differently all round), and also more == to === changes
* r10440 [2.1] Moving a function from Subs-Auth to Subs-Compat is great but it's not relevant to us since it's in 4.3.2 and up!

Notes
* r10404 Sinan also added a hook to ManageSettings.php for general mod settings but that was already added by Nao under a slightly different name. In SVN log: [SiNaN] knows that he'll be shot for doing this but he still has the heart of a modder.
* r10405 [2.1] also covers making queryless URLs work on Litespeed but that doesn't apply to us. There's also a removal of a @ error suppressor in the install file and a few == to === conversions. Really though, this is pure pedantry.
* r10408 implements it with browser conditions in the definition of $tag['content'] in the original array. Since that's not an option for us, I handled it inside the validate function. I did also fix the rather frustrating issue of IE not putting treating br tags as line breaks for copying, this is something they haven't.
* r10414 is a strange one. I'll buy validation of the URLs to prevent oddities (QueryString, Subs) but it also has two other changes: it removes the char 38 replacement in un_preparsecode, which seems to be OK because & is still escaped to its entity form (though I don't see what prompted it to be changed) and there is more robust checking of password flooding, specifically that a user isn't logged out when there's a password attack.
* r10439 The idea is to move the 'logging boards read', so that if goback is set, we fold that into the 'logging unread' process. Their method involves having an if and two queries' code, mine just manipulates the query variables, heh.
----
Modified : /trunk/Sources/Admin.php
Modified : /trunk/Sources/Class-Editor.php
Modified : /trunk/Sources/Display.php
Modified : /trunk/Sources/Dlattach.php
Modified : /trunk/Sources/Load.php
Modified : /trunk/Sources/ManagePermissions.php
Modified : /trunk/Sources/ManageSettings.php
Modified : /trunk/Sources/Post2.php
Modified : /trunk/Sources/QueryString.php
Modified : /trunk/Sources/Subs.php
Modified : /trunk/license.txt
Modified : /trunk/other/install.sql
Modified : /trunk/other/upgrade.php

Sadly not, but it is in Chrome which is how I wrote that script ;)

Antechinus said something interesting - and useful.All it needs is the CSS3 resize property set to 'vertical'.I took a few minutes out and stretched my brain, was quite nice.I don't write GM nearly enough.

Well, exchanging == for === in maintenance operations that happen sporadically if ever seems like a waste of time and effort to me. Doing that in heavy duty operations like parse_bbc can make a difference but in the 'empty unimportant logs' operation?

It doesn't help that only stuff in trunk appears in TortoiseSVN so if I want to see what's going on I have to look at it via their SVN viewer with its godawful 'view changes since previous' code.

Let's be completely fair for a moment: any environment where there are people who are paid a wage to provide you support should be able - all other things being equal - to provide you better support than a volunteer environment. It's a given, and one that too many on sm.org don't want to accept.

For a long time I've lamented about those who want the power of vB without the price-tag, though not quite for the same reason as you've mentioned.

Most of the mods on sm.org are not written by professional/semi-professional grade programmers, and when you apply changes at code level, it's simply going to become more and more fragile. This isn't me bashing SMF, this is a simple fact of life.

The ecosystem is crumbling because the essential requirements - enough people with the skills, the time, the energy - just aren't there any more. Even though I know for a fact there is now a branch in their SVN for 2.1, with half a dozen commits or so, they're all micro-optimisations rather than anything actually useful.


I will not get on a soapbox and preach and pretend that Wedge will be any better in the long term, because I don't know that. But I *do* know that we're distinctly moving towards a more mature codebase that makes it easier to write portable mods that don't edit the core code, that have no real reason to interfere with each other and so on.

I half expect we'll end up with the same shit attitude as on sm.org of people wanting it done for them, but I KNOW that a large proportion of the requests - namely "can we get this upgraded for x version" will really not be that much of an issue because mods that can cope with being portable just won't have that issue.