I developed a system like this for a previous job. We started with a source browser and then began checking engineering documents into our repository and (thus) making them browsable on the web
It can be really be nice. The hardest things were social issues; it requires a bit more discipline to maintain documents in a repository than by file sharing and email. Establishing the taxonomy so that people know where to put things and where to look for them is critical; but getting a good search engine up[1] can help there. Using small documents to redirect browsers to the "right" place can also help, since you can take a document that spans categories and put it in multiple places. I also wanted to get a Wiki in place, but left before we got there.
Having the revisions of design docs and test specs (which always change as implementation proceeds) tied to the appropriate revs of the software and other documentation using CVS tags, perforce change #'s/labels, or whatever is really nice.
As far as linking, HTML and MSOffice links were always made relative to the current document, so you could browse them when you'd checked them out on to your local machine. That was another social issue: educationing people to be careful, since most HTML editors and MSOffice HTML manglers of that vintage needed a little extra care to create relative links.
Since this was software engineering, we also implemented in-browser diffs (a lot like CVS web) and source code browsing as research tools. Having the design and discovery ("hey! so that's how this code works!" "Quick write it down before you forget!") docs that can link to the source code means that a new team member can read through a design doc and jump in and see the source code (where we kept implementation documentation).
You can modify your server-side source code browser to find things like file:../../design/index.xml in source files and convert them http: links when browsed, then use an editor or xterm that activates embedded links and be able to refer to design docs in soruce code. This also allow developers to link out to web sites, though we slurped anything that important in to the repo to protect against bit rot.
Since you've choosen XML a priori, you should definitely look at AxKit (Perl-based) or Cocoon (if Java's your fave) as delivery vehicles. Both are ASF official projects, though AxKit is a recent addition and hasn't made it to an Apache webserver.
Never used Cocoon, but AxKit can easily back-end to CVS and apply whatever transformations you like. AxKit can apply it's own XPathScript and XSP (language="perl") style sheets; various C-based XSLT engines (Sablotron and libxslt come to mind); and, of course, 100% Pure Perl to thoroughly munge your docs. Then it caches the results (if you like) and optionally GZips them (which is nice for dial-up or VPN use).
AxKit's main drawback at the moment is that it's web site is down due to sluggishness of British Telecom in installing a new data line, but you can find out more by searching cpan for the AxKit module.
If you do it right, you will have a very cool system.
HTH,
Barrie
[1] We did this before XML was all the rage, and getting meaningful searches of MSOffice files checked in to a reposiroty was a right pain. We ended up sharing out a directory tree that was updated nightly with the head revisions and letting people search with MSOffice's built in File Find.
They're working on that, they told me 6 months ago to sign up for announcements about their "server products".
Haven't heard anything since, though.
- Barrie
A fellow named Malcolm Beatty is/was working on a project called SILK (Simultaneuos Instances of the Linux Kernel or some such). Haven't heard anything about it in a long while though.
- Barrie
Microsoft has the right to ask you to remove materials that they hold the copyright to from your site unless you have paid for the right to put them there. However, that does not mean you need to remove the comment: merely substitute a message like
Check out Wiki-Wiki and some related works for a simple "text whiteboard" approach. Also, there's CVSWebEdit.
These are low-end systems, but they are in fairly wide use.
A related subject is collaborative annotation. this paper has a good review of tools, and CritLink is interesting.
I really want to work on this as part of an Open Source developer groupware app I'm working on, but the tuit supply is remarkably scarce at the moment...
Stats: CMU grad, worked here consulting for a few years, moved away, moved back.
Having lived here off and on since '83, I can honestly say I like "da burgh". There're good normal and more adventerous restaurants, mainstream and alternative cinema, good Rails To Trails, really excellent public theatre (Pittsburgh Public Theatre) and the Pittsburgh Symphony (site done by a friend of mine, FWIW), low traffic compared to other metro areas (Philly, NYC, Baltimore, and LA from direct experience), nice people, good high speed access in a growing number of neighborhoods, etc., etc. There's also an advocacy group of non-codgers trying to work on relevant issues for geeks and other young professionals.
On the political scene, the local county (Alleghany) just switched from a three-headed-dog-of-county-commisioners system to a single county executive. UPitt, CMU, Pittsbugh, and Alleghany County are starting to cooperate better in attracting business and supporting spinoffs and startups. CMU's actually figured out that the wild-eyed innovator rarely makes the best startup CEO and is trying to support startups/spinoffs with more serios business support. Pitt's learned that lesson as well.
Now that's not to say that we don't have 'issues', like tax structure, crappy roads, a high codger factor, but things are definitely on the mend around here. We haven't gotten to the point of choking on our own success: housing's, food nad clothing cheap, traffic's low (relatively), no more choking pollution. We've got good and growing support for ADSL (which brings you this missive), cable modems, and CPDP support.
All said, I'm glad I moved here and it looks like things are going to be moving in the right direction very nicely over the next 10 years or so. Meanwhile, we don't need to cope with the crap you need to cope with in longer-standing "high tech" areas.
Oh, and the standard disclaimer: "just my 0.02 worth".
I work for a large company selling an Oracle on SCO solution.
Since they're forcing us in to a replace, but not upgrade from OpenServer to UnixWare, we're taking the opportunity to evaluate Linux. Let's just say that the baroqueness of SCO's installation, upgrading, and licensing schemes have us strongly considering Linux. I suspect that Linux might out perform UnixWare 7, too. Not that I'm pushing it internally at all, of course:-).
All that being said, I really like SCO supporting linux apps, since it should encourage more people to develop for linux first, then let the SCO compatibility layer handle the 'port' to SCO.
Now, if only TokenRing under Linux is more stable than TokenRing under OpenServer 5...
Slashdot Mirror
I developed a system like this for a previous job. We started with a source browser and then began checking engineering documents into our repository and (thus) making them browsable on the web
It can be really be nice. The hardest things were social issues; it requires a bit more discipline to maintain documents in a repository than by file sharing and email. Establishing the taxonomy so that people know where to put things and where to look for them is critical; but getting a good search engine up[1] can help there. Using small documents to redirect browsers to the "right" place can also help, since you can take a document that spans categories and put it in multiple places. I also wanted to get a Wiki in place, but left before we got there.
Having the revisions of design docs and test specs (which always change as implementation proceeds) tied to the appropriate revs of the software and other documentation using CVS tags, perforce change #'s/labels, or whatever is really nice.
As far as linking, HTML and MSOffice links were always made relative to the current document, so you could browse them when you'd checked them out on to your local machine. That was another social issue: educationing people to be careful, since most HTML editors and MSOffice HTML manglers of that vintage needed a little extra care to create relative links.
Since this was software engineering, we also implemented in-browser diffs (a lot like CVS web) and source code browsing as research tools. Having the design and discovery ("hey! so that's how this code works!" "Quick write it down before you forget!") docs that can link to the source code means that a new team member can read through a design doc and jump in and see the source code (where we kept implementation documentation).
You can modify your server-side source code browser to find things like file:../../design/index.xml in source files and convert them http: links when browsed, then use an editor or xterm that activates embedded links and be able to refer to design docs in soruce code. This also allow developers to link out to web sites, though we slurped anything that important in to the repo to protect against bit rot.
Since you've choosen XML a priori, you should definitely look at AxKit (Perl-based) or Cocoon (if Java's your fave) as delivery vehicles. Both are ASF official projects, though AxKit is a recent addition and hasn't made it to an Apache webserver.
Never used Cocoon, but AxKit can easily back-end to CVS and apply whatever transformations you like. AxKit can apply it's own XPathScript and XSP (language="perl") style sheets; various C-based XSLT engines (Sablotron and libxslt come to mind); and, of course, 100% Pure Perl to thoroughly munge your docs. Then it caches the results (if you like) and optionally GZips them (which is nice for dial-up or VPN use).
AxKit's main drawback at the moment is that it's web site is down due to sluggishness of British Telecom in installing a new data line, but you can find out more by searching cpan for the AxKit module.
If you do it right, you will have a very cool system.
HTH,
Barrie
[1] We did this before XML was all the rage, and getting meaningful searches of MSOffice files checked in to a reposiroty was a right pain. We ended up sharing out a directory tree that was updated nightly with the head revisions and letting people search with MSOffice's built in File Find.
And when we ship prisoners off planet to small colonies and guard them with these giant roving balls....
Pennyfarthing bicycles on Mars anyone?
http://sources.redhat.com/glibc/glibc-faq.html
20-50 Hz, eh? So all those kids cruising in boomers must have really strong bones...
They're working on that, they told me 6 months ago to sign up for announcements about their "server products". Haven't heard anything since, though. - Barrie
A fellow named Malcolm Beatty is/was working on a project called SILK (Simultaneuos Instances of the Linux Kernel or some such). Haven't heard anything about it in a long while though. - Barrie
You might also check out tha AxKit, if you're thinking of moving towards XML+template processors.
Here's a backgrounder & introduction by the author (Matt Sergeant).
- Barrie
These are low-end systems, but they are in fairly wide use.
A related subject is collaborative annotation. this paper has a good review of tools, and CritLink is interesting.
I really want to work on this as part of an Open Source developer groupware app I'm working on, but the tuit supply is remarkably scarce at the moment...
- Barrie
Having lived here off and on since '83, I can honestly say I like "da burgh". There're good normal and more adventerous restaurants, mainstream and alternative cinema, good Rails To Trails, really excellent public theatre (Pittsburgh Public Theatre) and the Pittsburgh Symphony (site done by a friend of mine, FWIW), low traffic compared to other metro areas (Philly, NYC, Baltimore, and LA from direct experience), nice people, good high speed access in a growing number of neighborhoods, etc., etc. There's also an advocacy group of non-codgers trying to work on relevant issues for geeks and other young professionals.
And to note a few success stories: Fore Systems (now Marconi), Lycos, Free Markets Online (woulda liked to have a piece of that IPO), the Seagate magnetics research center, a mysterious whiz-bang startup, etc.
On the political scene, the local county (Alleghany) just switched from a three-headed-dog-of-county-commisioners system to a single county executive. UPitt, CMU, Pittsbugh, and Alleghany County are starting to cooperate better in attracting business and supporting spinoffs and startups. CMU's actually figured out that the wild-eyed innovator rarely makes the best startup CEO and is trying to support startups/spinoffs with more serios business support. Pitt's learned that lesson as well.
Now that's not to say that we don't have 'issues', like tax structure, crappy roads, a high codger factor, but things are definitely on the mend around here. We haven't gotten to the point of choking on our own success: housing's, food nad clothing cheap, traffic's low (relatively), no more choking pollution. We've got good and growing support for ADSL (which brings you this missive), cable modems, and CPDP support.
All said, I'm glad I moved here and it looks like things are going to be moving in the right direction very nicely over the next 10 years or so. Meanwhile, we don't need to cope with the crap you need to cope with in longer-standing "high tech" areas.
Oh, and the standard disclaimer: "just my 0.02 worth".
- Barrie
I work for a large company selling an Oracle on SCO solution.
:-).
Since they're forcing us in to a replace, but not upgrade from OpenServer to UnixWare, we're taking the opportunity to evaluate Linux. Let's just say that the baroqueness of SCO's installation, upgrading, and licensing schemes have us strongly considering Linux. I suspect that Linux might out perform UnixWare 7, too. Not that I'm pushing it internally at all, of course
All that being said, I really like SCO supporting linux apps, since it should encourage more people to develop for linux first, then let the SCO compatibility layer handle the 'port' to SCO.
Now, if only TokenRing under Linux is more stable than TokenRing under OpenServer 5...