test please delete

[NickFitz]

And as to how one represents all this in one's data model...

"This was the 150,000 post for about three minutes after which it became the 150,003 post once some new users were approved by the mods but then somebody (I'm looking at you, FG ) deleted several posts from ages ago and now it's 149,999"

[NickFitz]

Just been on a board where 8 bit character was zapped by the swear filter...

ROTFLMFAO...

UTF-8 FTW!

At the conference and unconference I've just been to there were speakers emphasising the importance of using UTF-8 everywhere, at all levels (database config up to specifying it in Apache config and HTTP headers)... I'd assumed that everybody already knew this

I really must stop assuming that everybody else knows all the bleeding obvious stuff - then I could get paid to speak at conferences

It was fun when we had our "bug bashes" at JustLeftClientCorp... one of my favourite tricks is to paste the umbrella from one of the Unicode pages full of dingbats into some text input field (password fields are good for this if people are watching you as they can't see what you've actually put in there). Often this was sufficient to stop the application working entirely and the hapless victim would have to bugger about at the MySQL command line trying to get rid of the record I'd created, without breaking the relationships to other records, while we all sat around and tutted

After that one could also try it with the G clef symbol, which lives in the area of Unicode known as the "Astral Plane"... the home of the "beyond 64K" glyphs

Still, as the point had normally been taken by then, I didn't usually bother.

Always a good test, though... there are ways of using malformed multi-byte UTF-8 characters to carry out SQL injection, XSS forgery, and all kinds of other security breaches on a web application - and believe you me, if I know how to do it, so do the bad guys

[BrilloPad]

morning

[BrilloPad]

Unemployment 1 %
Transport 99 %
Criminality 0 %
Pollution 1 %

pop=1075, rank=190

[BrilloPad]

Indeed...

That's one of the thorny issues I've been brooding about in relation to testpleasedelete.com - the fact that not only can posts be deleted, meaning that the TPD post number changes downwards, but posts by new users can also suddenly appear as if by magic (or moderation), meaning that the TPD post number changes upwards

Given that it could be the case that, say over a Bank Holiday weekend, a new user's post might sometimes take several days to actually become part of the thread (something that caught me out on a thousand around the 32K mark when DS24 became live - Hi FaQ ) it seems that there's a need to detect the up-bumping of posts (by realising that the unique CUK post ID no longer correlates with the original TPD post ID) and then crawling backwards, possibly for many pages, until one (or rather, the algorithm) finds the new post, after which one (or rather, the algorithm) can start sorting out the correlation between CUK IDs and TPD IDs.

And then seventeen (arbitrary number, but prime) new users get posts approved after an Easter weekend, so it's no good assuming that just one person has had comments approved over a short period - in principle it could take an enormous amount of working out

Which is of course why one devises an algorithm to do it... if the "everything is done by the algorithms" policy works for Google (who strive to do absolutely everything algorithmically) then it ought to work for us

IMO unapproved posts should be shown with "awaiting approval". deletion would be impossible (just update to "..").

I am sure the mods would love to help with the integrity of tpd!

[BrilloPad]

And as to how one represents all this in one's data model...

"This was the 150,000 post for about three minutes after which it became the 150,003 post once some new users were approved by the mods but then somebody (I'm looking at you, FG ) deleted several posts from ages ago and now it's 149,999"

Was that FG or was that a mod?

[BrilloPad]

UTF-8 FTW!

At the conference and unconference I've just been to there were speakers emphasising the importance of using UTF-8 everywhere, at all levels (database config up to specifying it in Apache config and HTTP headers)... I'd assumed that everybody already knew this

I really must stop assuming that everybody else knows all the bleeding obvious stuff - then I could get paid to speak at conferences

It was fun when we had our "bug bashes" at JustLeftClientCorp... one of my favourite tricks is to paste the umbrella from one of the Unicode pages full of dingbats into some text input field (password fields are good for this if people are watching you as they can't see what you've actually put in there). Often this was sufficient to stop the application working entirely and the hapless victim would have to bugger about at the MySQL command line trying to get rid of the record I'd created, without breaking the relationships to other records, while we all sat around and tutted

After that one could also try it with the G clef symbol, which lives in the area of Unicode known as the "Astral Plane"... the home of the "beyond 64K" glyphs

Still, as the point had normally been taken by then, I didn't usually bother.

Always a good test, though... there are ways of using malformed multi-byte UTF-8 characters to carry out SQL injection, XSS forgery, and all kinds of other security breaches on a web application - and believe you me, if I know how to do it, so do the bad guys

I think I need to attend one of your conferences!

[BrilloPad]

Unemployment 1 %
Transport 99 %
Criminality 0 %
Pollution 2 %

pop=1075, rank=190

[BrilloPad]

morning all

[Advocate]

morning all

Mornin'