OT: Knowledge management (for Embedded engineers)

The best one is sitting on your shoulders.

Ian

Yes. A set of (paper) logbooks, a big pile of box-files (some for projects, some for technical data) and a well-indexed mailbox (incoming and outgoing) works fine for me. Don't invent more complicated systems. And never delete email.

pete

My answer is to run a TWiki site locally, which allows me to easily structure a collection of links and cached original sources as web pages. I can also run a fairly sophisticated search on the contents. Details are here...

It's not the ultimate solution, but it works for me.

- Richard

I also recommend a Wiki, but not absolute TWiki and/or "locally".

I'm running PhpWiki (expecting a stable version 3.11 with access control) but other enginges should be considered: PMWiki is lean, MediaWiki rich.

A web based Wiki has the advantage that I can access it also from outside my office.

Oliver

I second this suggestion. A wiki works very well for me.

Kelly

I store my info in XML and HTML pages, but don't use a wiki for that purpose (though I have used them). I just tend to knock out either very simple HTML pages using a text editor (vim) or very complex XML pages with embedded SVG, for example, using the same text editor. I have my pages arranged something like this:

docs/ docs/catalogs ... docs/datasheets docs/datasheets/AnalogDevices docs/datasheets/AMD docs/datasheets/Atmel ... docs/manuals docs/manuals/software docs/manuals/equipment docs/manuals/tools ... docs/standards docs/standards/languages docs/standards/RFC docs/standards/ANSI docs/standards/IEC ...

I have another tree which contains more dynamic data (project data typically) which hyperlinks over to the docs/ tree. Using this, and a couple of Perl-based link-following tools, I can quickly extract out and archive a whole project (including standards, databooks, schematics, source code, etc) onto CD. The nice part about this is that I can hand the CD over to somebody who runs Linux, or Windows, or Mac OSX and everybody sees the same thing -- there is no executable content; only hyperlinked pages. Some of it looks pretty slick, e.g. two versions of a schematic. One in TIFF which is zoomable and scaleable (requires a TIFF viewer or plug-in) and the other is an HTML map which vectors you to the datasheet of the part when you click on a part on the schematic

-- sort of a visual BOM. However, the process by which I make those is still pretty manual.

The only thing I don't have is an indexer or search tool that will run cross-platform. Maybe a Java application would be useful for this purpose, but I've been getting along without one for a while now.

Ed

Hi Ed,

A long time ago one of the search engines offered a tool for local disk searching and indexing, maybe it was 'Altavista Personal' or something like that. Basically you got the same feel as if you did a web search. I believe the new Google Desktop Beta version is something similar and it could soon spread across platforms. If you want to check it out:

Regards, Joerg

IT

processes.

You wrote:

Just one question. How is the requirement for changeability handled? A canned solution can be very effective and solve all of your problems ... if and only if your problems are the set of problems envisioned by the authors of the collaboration solution ...

What is the approach to changeability and customizability?

This is not an argumentative reply. I'm curious what others have done in this area ...

Thanks, Dave.

Hi David,

It has been a few years ago but with Agile we were able to change the process as needed. For example, if a procedure used to be done via a deviation and we wanted to stiffen that up and require a change order that could be programmed in. Same when we wanted to incorporate some cost data that hadn't been required upon installation. Sure, sometimes that required getting the vendor's programmers out but much of it could also be done by our own IT staff.

The main thing I was looking out for in the pre-purchase meetings with the vendor or this enterprise system was how much we could customize without needing them on site. That was simply a cost concern. But as far as I remember there were not a lot of roadblocks in customization. Except that stuff that would violate a certified performance to FDA standards was not allowed, of course. We did have to customize in that area as well because we had to deal with foreign agency requirements and also overseas subsidiaries.

One thing I should point out here though is that these systems are meant to control the data that "has to be" controlled, plus some data that we wanted to be there. So if you needed to know anything about a product or a process and had the proper authorization the PC was all you needed to get that data. No more trips to the doc center (except when they were selling girl scout cookies). But these systems are not really suited to casually collect information as people come across it. That would be better handled with a Wiki solution.

Regards, Joerg

You are not alone for sure. We all have various ways to manage the wealth of information, data and documents as well as managing a multitude of versions and the attendant problem reports. There are some good, if expensive, commercial products out there. In one large project I managed to persuade them to install a combined version control and problem tracking system that flows the way my own process does. It was worth the $500+ per seat to them as they had a very large development team spread globally. Even so, there were a couple of slight shortcomings with that package for which we had to implement a manual procedure to cope with them.

Aside from improving the integration between existing tools what would you suggest could be done to assist us with this information overload and scheduling. Personnaly I consider that adopting a few, widely accepted standard formats (HTML, XML, PDF etc), assists in the task of storing the information. You need to ensure that the tools you use can read these in (for modification) as well as generate them.

Agreed, version control and change management are very important issue and is probably one area where more work should be expended by the open community. Not just for software but for all documentation. An open source problem report tracker would be useful too, especially one that works with the open source RCS or VCS systems.

Agreed. That solution should also incorporate the means to automatically add the identity of the last person to change the details (and prompt for a reason for change). [%X]

Agreed. When many terminal solutions (be they PDA, PC or X-terminals) seem to be incorporating capable browsers as a matter of course. At the document management level browsers are the easiest way to initiate document searches and display.

Keep us posted. I am beginning to build a multi-machine server network here too and will be interested in what you do in this respect. Naturally I am implementing Linux as the OS and am currently revising the security aspects of the other machines that will perform the specific server tasks.

I don't suppose you know how much (per seat) Agile is. Their website didn't say. It looked quite similar to the MKS product range, one i have experience with (see

notes

the

Well, I'm using word documents and all my datasheets in PDF has been scanned by ABBYY Finereader (it will scan PDF and convert it to text). So all my data are in text format and a simple tool like Total Commander let me scan for a word through all the stored info. Perhaps there's a better way?

Cheers

Klaus

Hi Paul,

It's too long ago but it was a major purchase since we had dozens of seats. The cost will greatly depend on the level of compliance you need. In our case we had to be totally FDA compliant and that, of course, adds cost for extra customization, validation, certification and so on. Our other main concern was to make sure that such a major revamp didn't bring us all to a screeching halt. We had a fully running business and even a day of turmoil would have been pretty traumatic. It went smooth.

Also, training costs factor in and if you have 50-100 people who must become proficient on a new system that can be a large chunk of the expense. So even if they gave a generic cost per seat this could not be much more than a "guesstimate".

The nice thing about this and other systems that are pretty all-encompassing is how fast you get to information. Once I had an idea and wanted to see if another division could use that in one of their products.15 seconds later I had their schematic on screen. Almost that minute the phone rang and there was a problem with a particular chip in our production. While on the phone and within a few seconds I had the stock room qty and future ship schedules right on my screen. All via one system. What took me a total of five minutes used to be at least a half hour eventhough I must say that the half mile total walking distance used to be healthier than having it all on the screen. Except when the doc center offered those mint cookies.

Regards, Joerg

Well, to answer my own post - just found this:

/Klaus

Yes! I still have a copy. You can get it to index all the files on a LAN by mounting all the remote computer drives "on" your PC!

or something

If you don't want to spend 200USD (or incredible 277EUR for a German version), try

Not so powerful as dtSearch but worth the money (also worth the time). Indexing was slow, but the index was small and search results appeared almost instantaneously. Preview for Word, Excel, PDF... - really nice freeware.

Maybe

Oliver