hi I last played around with Zope maybe 18 mos. ago and didn't particularly care for it, but have recently gotten interested again- either it's improved greatly or I have or maybe both, but in any case I'm seriously considering it for use in a rather peculiar project and would appreciate any suggestions that you may care to offer. Briefly, a client of mine has an intranet web site that contains its ISO-9000 documentation. The content is almost entirely in native Word, Excel, etc. format (not because that's a good idea, it's just the way things had to be and have to remain as a practical matter). There are over 1600 documents and around a third of them have 2700 hyperlinks embedded in the text flow that use relative addressing. As you would expect, the file sizes are huge- the largest is 9.4 megs and the average is 181k- if you throw out the largest 50 and smallest 50, the average is still 114k. Anyway, this has worked better than it should for about a year in spite of being served by the miserable IIS on NT and I've used python scripting to handle the menu maintenance and to do some fairly arcane reporting. The site is actively used by maybe 150 people a day and something significant changes at least 2-3 times a week. Suffice it to say that there have been regular problems as well, about 80% because of IIS (i.e., mystery file and folder locking for weeks at a time and other functionality that you can read about in the M$ bug database) and the balance because the IT guys are slackers who apparently have very little understanding about how NT works. The upshot is that now there is talk of migrating the whole thing into Lotus Notes (using a web interface) and actually storing the files in Notes mainly to obtain better reporting. This strikes me as absurd given the file sizes and the need to reset all of the embedded links in the documents among other things, so I am thinking more along the lines of a) getting this site moved onto its own server, preferably either bsd or linux, and b) using apache to serve the static files from the file system and c) using Zope to handle the interface, any DB connections that are appropriate, and managing things via ExternalFile. I would assume that the internal links will work without any changes since the file system is still the same. Does anyone see a big problem with this approach? And does anyone have experience with Lotus Notes in this type of project? I've never heard anything good about Domino server except that it was appropriately named, but.... Thanks & sorry to ramble on for so long. Richard Gordon -------------------- Gordon Design Web Design/Database Development http://www.richardgordon.net
[Richard Gordon]
... Does anyone see a big problem with this approach? And does anyone have experience with Lotus Notes in this type of project? I've never heard anything good about Domino server except that it was appropriately named, but....
Our company uses Notes/Domino on its internal and external web sites - although I haven't worked with it myself - and it handles many more files of all kinds than you mention without apparent trouble. The Notes database, I am told, has both relational and object features, and seems to be very capable. I have no idea how it would handle the relative links in the documents. Zope might well be a better solution, but Notes/Domino could be a real contender if there is a way to handle the link problem. It would take some technical analysis, which might include the ability to keep using your Python reporting stuff with minimal changes (a plus for Zope) vs whether there is strong inhouse expertise in Notes. Cheers, Tom P
At 4:36 PM -0500 1/30/02, Thomas B. Passin wrote:
Our company uses Notes/Domino on its internal and external web sites - although I haven't worked with it myself - and it handles many more files of all kinds than you mention without apparent trouble.
Thanks for the reply. I wasn't concerned with the number or kinds of files so much as their weight. IIS is already groaning sometimes.
The Notes database, I am told, has both relational and object features, and seems to be very capable.
hmm. I understood that it essentially had neither. http://www.woomeranet.com.au/bazaar.enterprise.html
Zope might well be a better solution, but Notes/Domino could be a real contender if there is a way to handle the link problem. It would take some technical analysis, which might include the ability to keep using your Python reporting stuff with minimal changes (a plus for Zope) vs whether there is strong inhouse expertise in Notes.
The link problem is pretty bad. I've never found anything in the Word api that will let you automate global changes to hyperlinks via regular expressions and, even if you could, I found that a huge number of them are essentially corrupted by the ways that Word stores changes (in other words, an active version works, but the document can be littered with artifacts). Anyway, there is no inhouse Notes expertise- I am up against some other contractors and do have the dubious advantage of working for about half their hourly rates. Why Notes people can charge more than Oracle and Java programmers is not clear to me. Anyway, thanks again. Richard Gordon -------------------- Gordon Design Web Design/Database Development http://www.richardgordon.net
Does anyone see a big problem with this approach?
None whatsoever. Although I would try to move everything over to storing everything in native HTML directly in the Zope DB and using a content management system for direct editing of the content by the users. But thats probably just me. :-) And even if it's not, doing things one step at at time is probably a good idea.
And does anyone have experience with Lotus Notes in this type of project?
Only from customers throwing it out in favour of Zope + a CM system. Well, OK, one ustomer... :-)
At 1:17 AM +0100 1/31/02, Lennart Regebro wrote:
None whatsoever. Although I would try to move everything over to storing everything in native HTML directly in the Zope DB and using a content management system for direct editing of the content by the users.
Conversion would be preferable, but it's not going to happen. There are several reasons, but the leading one is that they insist on concocting heavily formatted and illustrated Word documents that just don't render well in html. Even if you can manually overcome the deficiencies of the various conversion utilities available [I exclude the Zope product from that statement for now because I haven't actually tried it, but everything else certainly sucks], what they really want is what amounts to a picture of the file, but they hate pdf, so.... I've resigned myself to having to contend with Office documents and all of the many, many pains associated with them. On the bright side, this nightmare did present me with an opportunity to use python to write the only linkchecker that I know of that will test links embedded within Word documents and if I ever get around to cleaning up the code so that it is suitable for general use, I'll probably open source it. Thanks for the reply. Richard Gordon -------------------- Gordon Design Web Design/Database Development http://www.richardgordon.net
participants (3)
-
Lennart Regebro -
Richard Gordon -
Thomas B. Passin