I was asked on Friday night if I could make a demo foswiki USB system, and given that its high time we did one, I started to look into it. Initially I thought this would be a great oportunity to try out the HTTPEngine work Gilmar has been doing, but we’re not quite there yet. And so, I started a quick perusal of the existing WebServer on a Stick systems. Ideally, I want to use Strawberry perl, apache, and have the server and a browser start up automatically when the USB stick is inserted into the computer.

• MicroApache
• Server2Go
• XAMPP

MicroApache was the first thing i looked at, as the contact pointed out DokuWikiOnAStick, but that server doesn’t seem to have source, nor is the upstream web site there (last release seems to be 2007, so too old for Strawberry perl too) Server2Go looks nice, but as its not really free in the debian sense, I’ll pass on that unless nothing else works. XAMPP – unzipped it, foswiki and did some minor configurations. unziped, removed everything except apache, added strawberry perl, and wrote a systray icon and menu system for you to control it with. TADA! FoswikiOnAStick v0.1v0.3 (running foswiki 1.0.6). please try it out.

FoswikiOnaStick system tray menu

Instructions for use:

1. download the Foswiki on a Stick zip file
2. unzip it to the root of your USB key (or any drive)
3. start the systray app, web server and browser using start_foswiki.bat (goto http://localhost/ if the browser doesn’t start) (unblock / allow apache to get through the windows firewall)

before ejecting the usb key, you need to stop the webserver by running stop_foswiki.bat yes, thats all. Please report any issues or observations to me – SvenDowideit@fosiki.com

• the web server runs at the standard port 80, so if there is already a web server running it will fail
• when the web server starts, it will probably be blocked by Windows firewall – every time I’ve tested, a dialog pops up, and allows you to unblock it.
• ….

Warning

I have made no attempt (yet) to reduce the accesses to the disk – so it may use up all the blue smoke in your usbkey, cause your bits to turn green, or one of many other unforseen side effects which may or may not appear amusing if they happen to someone else. I’ve tested this on Windows7rc, Windows Vista, Windows2003server, and Windows XP – if it works for you, excellent .

Yesterday I uploaded the first Foswiki Installer for windows to Foswiki Downloads. It will install Apache 2.2 (if you don’t already have it) and Strawberry Perl and then configure and install Foswiki v1.0.0 beta 3.

In a clear example of TWiki.NET’s new no-testing policy [1], Their star worker has yesterday un-necessarily uploaded over 300 plugins from the wrong branch in Subversion.

Rather than testing each contrib before uploading, as has been the policy on TWiki.org since the Plugins web was created, TWiki.org is now full of buggy, outdated or just plain wrong packages. The only stable option they have is to revert the uploads using rcs – as many plugins are either not up to date in subversion, or worse, are works in progress in Subversion.

Adding that to the latest TWiki 4.2.4 release, which was released with failing unit tests, and some very dubious changes:
we have already submitted to them an updated exploit that is present in 4.2.4 for the CVE-2008-5304 XSS exploit
- presumably they will release 4.2.5 in the next week?

Its a pretty sad time for those of us with users on TWiki (as Foswiki hasn’t released yet), but it has certainly spurred us along, now we need it more than we ever expected.

[1] Tom Barton: issues … impeding progress: an excessively rigorous approach to testing that actually inhibited less experienced developers from contributing code; …

if you’re working or just running Perl on Windows, drop everything, run, don’t walk, to StrawberryPerl. Adam Kennedy has not only made a real Perl for windows, he’s made a _proper_ Perl . One where CPAN just plain works.

Even better, he’s made a Perl that you can use portably, from your USB stick, so you don’t even need to install Perl on your locked down computer.

To learn about his code, I’m building a FoswikiOnAStick distro based on his code, and then I hope to work out how to extend the concept to other platforms.

CPAN, while incredibly useful, can be a pain, if you forget that you need to re-configure it after installing essential tools.

For example, if you make the mistake of setting up a basic, non-development Debian virtual machine, configure CPAN, try to use it, and on seeing ‘make’ errors like (from install Bundle::CPAN of all things) :

Running make test
Can't test without successful make
Running make install
make had returned bad status, install seems impossible
Running install for module Compress::Raw::Zlib
Running make for P/PM/PMQS/Compress-Raw-Zlib-2.012.tar.gz
Is already unwrapped into directory /root/.cpan/build/Compress-Raw-Zlib-2.012
Has already been processed within this session
Running make test
Can't test without successful make
Running make install
make had returned bad status, install seems impossible
Running make for P/PM/PMQS/IO-Compress-Zlib-2.012.tar.gz
Is already unwrapped into directory /root/.cpan/build/IO-Compress-Zlib-2.012
Has already been processed within this session
Running make test
Can't test without successful make
Running make install
make had returned bad status, install seems impossible

cpan>


You install make apt-get update ; apt-get install build-essential…, only to continue to see the same errors wizz past….

CPAN really truly needs to realise that the make settings are mis configured, and tell you.

What you need to do, is to tell your cpan about it by running:
cpan> o conf init

OR, if you’ve not yet messed (configured) up your cpan, install build-essential first.

And while you’re contemplating using cpan, think hard about trying dh-make-perl instead

Ideally, CPAN should be able to realise that it can’t call make if it does not know where it is – and point this fact out, rather than making it appear as though the package being installed has an issue.

I’ve just updated the Experimental TWiki and Plugins repository. It now contains TWiki 4.2.0 and 212 Plugins, Contribs and Skins that you can simply apt-get install

To use them, add the following 2 lines to your /etc/apt/sources.list

deb http://distributedinformation.com/experimental/ experimental main contrib
deb-src http://distributedinformation.com/experimental/ experimental main contrib


then type

apt-get update

to update the available packages.

you can now see all 212 packages with apt-cache search twiki-

and install (assuming you don’t have twiki installed yet)

apt-get install apache2 twiki

and TWiki Contrib installation is as easy as

apt-get install twiki-bugscontrib

You will still need to use configure to enable Plugins.

Please report your experiences to me – bugs, gripes, you name it – its a work in progress. and I need your help!

I was having a discussion with someone on IRC about how TWiki is vulnerable to Cross-site scripting and Cross-site request forgery, and we realized that there are 2 possible approaches to securing TWiki effectively (both requiring a unique magic number for all URLs):

1. add a pre process to the TWiki::UI system, requiring a valid and unique magic, and a post process step between rendering and output to the browser
2. use a small proxy system between TWiki and browsers to add and validate the magic

1. is actually still risky as all scripts still are able to output directly to the browser using a =print= statement, thus giving the user urls that may not have a necessary magic in the url, or similarly for AddOns that persist in not using resthandlers.

whereas 2. abstracts the security from the application server, in much the same way as it is for ssl – goodness all round.

So – I wonder if there is such a proxy already?

There are also massive performance reasons why you should always have a proxy between browsers and heavy application servers like TWiki – this too could do with filling out.Securing TWiki is not as simple as converting all actions to POST (ie using proper REST / HTTP) because there are too many legacy conveniences, allowing GET URL’s to act upon the data. But, by delegating the securing of the transactions to an external wrapper, I think we can avoid these flaws.
see Wikipedia on Cross Site Scripting and Cross-site request forgery

I have just uploaded the OpenIDUserContrib for TWiki 4.2.0. It adds OpenID login and 1.1 Attribute functionalty to TWiki.

Currently, it disables Registration, and limits authentication to OpenID users.

It has the advantage over the OpenIDAuth apache module, that it automatically requests the User’s OpenId 1.1 attributes like Name, Email address directly from their OpenID identity.

While it trusts the user’s choice of ‘FullName’ registration attribute when displaying who made changes to topics, the TWiki topic source
actually stores the authenticating OpenID URI, thus their user details will be updated from the authentication server next time they log in.

Note that TWiki Topic based Groups are not yet implemented using this Mapper.

Future directions

• add mixing of UserMappers to allow OpenID and ‘normal’ TWiki auth
• turn TWiki into an OpenID identity server
• add Safe Group definition system
• add OpenId to TWiki’s registration process (would require openid auth first, then prefill registration details from any available attributes
  • This will require re-writing of TWiki’s inbuilt registration system
• move the list of Known users and their mapping information from data/OpenIdUsers.txt to somewhere more scalable. (perhaps DBI)
  • combine the info TWiki uses persistently with the Session and other caching info OpenID11? uses

Net::OpenID11 (based on Net::JanRain::OpenID)

To make this work, I fixed the Perl bugs I found in Net::JanRain::OpenID, and renamed the resulting modules under Net::openID11 (as it is not OpenID2.0). I expect to upload these packages to CPAN some time soon.

If you want to take a look at the code – goto my Subversion repository

These Windows, OSX, Centos and Fedora Core installers are fully integrated native installers that will update your Computer with perl, apache, rcs and other tools needed to run TWiki on that platform.TWiki 4.2.0 contains many new improvements to TWiki, including a much improved Wysiwyg editor, a structured query engine, a more generic authentication system and at the same time, the Core engine is faster than the previous twiki4 releases.The TWiki installers include native installs of (only installed if not already)

1. Apache 2.2 (Windows & rpm)
2. Perl (ActiveState – Windows & native for rpm)
3. Gnu Grep (Windows only)
4. Gnu rcs (All platforms)
5. TWiki 4.2.0 Release.

Please download it, try it out and report your impressions, ideas, bugs and successes here, on TWiki.org, or in the TWiki Bugs system.

• Windows TWiki installer 4.2.0  (23MB)
• Mac OSX TWiki installer (Tiger only I don’t yet own Leopard – donations welcome) 4.2.0  (2MB)
• Centos 4&5 and Fedora Core 5&7 4.2.0  (17MB)

Another TWiki innovation brought to you by distributedINFORMATION & WikiRing.com

For the extremely adventurous – I have built an installer using Strawberry Perl 5.10 beta2 – TWiki-4.2.0-rc2.1-strawberry.exe

Warning: Search does not work, and needs someone to debug it (I’m away over xmas)