Hi; Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours. What inspired this post was something that happened tonight. I was playing around with a very simple script when I kept running into errors: Can't convert an integer into a string; Can't convert a string into an integer; etc. I *knew* there wasn't a problem with my script. After a futile hour of trying to track down the *problem*, I decided to copy a perfectly good and currently functioning script, giving it a different name and passing the variables like I do to its parent. Sure enough, I ran into the same errors! Why?! My worst nightmare happened a couple of weeks ago. I decided to rebuild my file structure into something more portable. I had placed a number of scripts and old DTML methods in the root. I decided to copy them all into a folder, add the folder to each of my sites, and put the new folder into each site's path. Simple, right? What could possibly go wrong? Well, I forget the exact errors Zope threw at me, but it was all utter nonsense. I finally decided I'd just restart the sucker. I mean, why not? Bad, bad idea. I couldn't restart it. Worse yet, I ended up having to reinstall Zope. When I went to open my old Data.fs file, the goblins had secretly returned it to the exact state it was in at my last backup...two weeks before! How???? How could it have possibly known?! Also, if you happen to know a good exorsict, please let me know <g> beno
Have many stories about The Goblins. For example, Permissions, not working in *particular* folder. I run http://www.xmlsuck.com/ on Squishdot for 1+ year. One day, all of a sudden, 'Search' stopped working. Manage->ControlPanel->pack has fixed it. In fact, it was not *that* simple, because after 'fixing' it that way, Zope has crushed several times next 48 hours, but I was stubborn and now, couple weeks later, everything works fine. I belive that : - everything in Zope depends on ZODB - ZODB is not rock solid So - any day you may start experiencing some funny effects. Not frequently. I would say - once two months, maybe. "ZODB is not rock solid" needs more comment. ZODB has pretty nice self-recovering abilities. In fact, I was *very* happy (and was not expecting) that the 'pack' thing would be able to 'fix' ZODB - from the shape it was. Basically, ZCatalog and everything depending on it was not working (not only xmlsuck part), so I was expecting the damage to be major. I'm making backups, but still ... Something like "Norton Disk Doctor" for Zope may make sense eventually, when/if Zope would gain more marketshare. Perhaps there is some utility out there which is devoted to capturing / fixing inconsistensies in ZODB 'at all costs' ? Some improved version of Manage->ControlPanel->pack ? I tried looking, but found nothing comparable to "Norton Disk Doctor". Rgds.Paul.
At 07:26 PM 1/12/2003 -0800, you wrote:
I belive that :
- everything in Zope depends on ZODB - ZODB is not rock solid
Bingo! Yep: me too.
So - any day you may start experiencing some funny effects. Not frequently. I would say - once two months, maybe.
"ZODB is not rock solid" needs more comment. ZODB has pretty nice self-recovering abilities.
In fact, I was *very* happy (and was not expecting) that the 'pack' thing would be able to 'fix' ZODB - from the shape it was.
Hmm. I just tried that (sounded like a good idea). Got a *permission denied* error. So I tried it again as *admin*. Same error. Not good. beno
Paul T \(spk\) wrote at 2003-1-12 19:26 -0800:
... I belive that :
- everything in Zope depends on ZODB - ZODB is not rock solid It is quite solid and it is *very* resilient to data loss, unless you pack it (and then you have the old unpacked file as backup).
... "ZODB is not rock solid" needs more comment. ZODB has pretty nice self-recovering abilities. Almost nothing can go wrong with ZODB and FileStorage:
Any changes are added at the end of the file (okay, Toby reported that ZODB provides features that allow modification in the middle, but Zope does not use them). You have the complete history, in case you must recover (or analyse). Desaster recovery means, you cut away some trailing part of the "Data.fs"
... Perhaps there is some utility out there which is devoted to capturing / fixing inconsistensies in ZODB 'at all costs' ? Some improved version of Manage->ControlPanel->pack ? There are "ZODB.fsrecover" (for recovery) and "ZODB.fsdump" and the third party product "tranalyzer" for analysis
I tried looking, but found nothing comparable to "Norton Disk Doctor". I work under Unix. Therefore, I do not know (and usually do not need) "Norton Disk Doctor"
Dieter
"ZODB is not rock solid" needs more comment. ZODB has pretty nice self-recovering abilities.
Almost nothing can go wrong with ZODB and FileStorage:
Unfortunately, something can. http://www.google.com/search?hl=en&lr=&ie=UTF-8&oe=UTF-8&q=Data.fs+problem In particular : http://www.zopezen.org/Members/macguyver007/1005847669 ( published at Sep. 27. 02 ) To me it sounds like one more argument for "daily packing is good for you".
Any changes are added at the end of the file (okay, Toby reported that ZODB provides features that allow modification in the middle, but Zope does not use them).
You have the complete history, in case you must recover (or analyse).
Desaster recovery means, you cut away some trailing part of the "Data.fs"
I'm not sure this is relevant to what happens with packing.
Some improved version of Manage->ControlPanel->pack ? There are "ZODB.fsrecover" (for recovery) and "ZODB.fsdump" and the third party product "tranalyzer" for analysis
Thank you for the pointers.
I tried looking, but found nothing comparable to "Norton Disk Doctor".
I work under Unix. Therefore, I do not know (and usually do not need) "Norton Disk Doctor"
You have it invoked for you at startup - transparently ( it is called fsck ). In fact, Windows does the same thing these days, so I think that many newbie Windows users would also not understand my reference to NDD. I should have said not 'Norton Disk Doctor', but "some utility that fixes the broken filesystems and does it *really* well". Perhaps a combo of all three things that you've mentioned + 'pack' would be 'it'. Rgds.Paul.
On Monday 13 January 2003 7:31 pm, Dieter Maurer wrote:
I tried looking, but found nothing comparable to "Norton Disk Doctor".
I work under Unix. Therefore, I do not know (and usually do not need) "Norton Disk Doctor"
If a unix filesystem did not have a competent "fsck" tool, would you use it in production?
Desaster recovery means, you cut away some trailing part of the "Data.fs"
That depends on how much you have to throw away. If the problem goes unnoticed too long then this is a "recovery disaster", not a "disaster recovery". How does anyone know for sure that there is no problem in their storages today without using a "fsck" tool? -- Toby Dickenson http://www.geminidataloggers.com/people/tdickenson
Toby Dickenson wrote:
On Monday 13 January 2003 7:31 pm, Dieter Maurer wrote:
I tried looking, but found nothing comparable to "Norton Disk Doctor".
I work under Unix. Therefore, I do not know (and usually do not need) "Norton Disk Doctor"
If a unix filesystem did not have a competent "fsck" tool, would you use it in production?
Would you use xfs? http://oss.sgi.com/projects/xfs/manpages/fsck.xfs.html ""NAME fsck.xfs - do nothing, successfully [...] """ Yes, it's not quite the same, but I think this manpage is very funny ;). cheers, oliver
On Tuesday 14 January 2003 10:29 am, Oliver Bleutgen wrote:
If a unix filesystem did not have a competent "fsck" tool, would you use it in production?
Would you use xfs?
http://oss.sgi.com/projects/xfs/manpages/fsck.xfs.html
""NAME fsck.xfs - do nothing, successfully
I wouldnt rule it out for not having a fsck tool, although giving standard tools silly readable names is a very good reason to shun it: http://oss.sgi.com/projects/xfs/manpages/xfs_check.html
Yes, it's not quite the same, but I think this manpage is very funny ;).
Indeed -- Toby Dickenson http://www.geminidataloggers.com/people/tdickenson
-----Original Message----- From: zope-admin@zope.org [mailto:zope-admin@zope.org]On Behalf Of Toby Dickenson Sent: Tuesday, January 14, 2003 2:18 AM To: Dieter Maurer; Paul T (spk) Cc: zope@zope.org Subject: Re: [Zope] The Goblins of Zope
On Monday 13 January 2003 7:31 pm, Dieter Maurer wrote:
I tried looking, but found nothing comparable to "Norton Disk Doctor".
I work under Unix. Therefore, I do not know (and usually do not need) "Norton Disk Doctor"
If a unix filesystem did not have a competent "fsck" tool, would you use it in production?
ZODB (as used by zope) is not a filesystem and, in fact, is a great deal simpler. Real FSs have to manage a fixed amount of space, which means they have fragmentation and free lists to handle. ZODB has no such issues, so saying it needs a fsck is much like saying /etc/password needs a good fsck. It doesn't. It's just a linear list of data records. As Zope uses ZODB, it too is pretty much just a linear list of objects. That's not to say it's bug free or flawless. But saying it needs a fsck facility is just not true. Lopping off the end is a fine way to roll back and a fsck wouldn't do any better.
Toby Dickenson wrote at 2003-1-14 10:18 +0000:
... If a unix filesystem did not have a competent "fsck" tool, would you use it in production? If the physical structure of the Unix file system were as trivial than that of FileStorage, then maybe...
... How does anyone know for sure that there is no problem in their storages today without using a "fsck" tool? You are much deeper in "Storages" than I am.
I have the feeling that a "FileStorage" is a linear sequence of transaction records. When Zope builds the "FileStorage" index, I expect (I never verified) that is analyses the linear sequence and checks that at least the record sizes are correct (otherwise, with high probability, invalid transaction records would be encountered). Besides this elementary structure, there are backpointers from newer to older versions. Thus, I expect, that damage to the record structure (modifying the length of a record) would be detected the next time when Zope starts. Damage inside a transaction record may not be detected. But, if the problem indeed, is unnoticed for a long time, then it does not seem to be that grave. Dieter
On Tuesday 14 January 2003 6:32 pm, Dieter Maurer wrote:
How does anyone know for sure that there is no problem in their storages today without using a "fsck" tool?
You are much deeper in "Storages" than I am.
I have the feeling that a "FileStorage" is a linear sequence of transaction records. When Zope builds the "FileStorage" index, I expect (I never verified) that is analyses the linear sequence and checks that at least the record sizes are correct
When building the index FileStorage scans through these headers as fast as possible. This will detect some damage but not all. Of course this check is only performed when the index is built - on startup after an unclean shutdown. During a clean shutdown the old index is persisted in data.fs.index. If you always shutdown cleanly then this may never happen!
Besides this elementary structure, there are backpointers from newer to older versions.
Yes, plus other redundant information. For FileStorage, all of this is thoroughly tested with the fstest.py script. IMO it is prudent to run fstest at least as often as you pack, perhaps on Data.fs.old.
Damage inside a transaction record may not be detected.
For FileStorage, damage at pickle level and ZODB level (dangling references etc) can be checked with fsrefs.py. For DirectoryStorage, checkds.py checks as much as it can, plus you can run it on a live storage. Im not sure about the situation for BerkelyStorages. -- Toby Dickenson http://www.geminidataloggers.com/people/tdickenson
On Tue, Jan 14, 2003 at 09:41:02PM +0000, Toby Dickenson wrote:
For FileStorage, all of this is thoroughly tested with the fstest.py script. IMO it is prudent to run fstest at least as often as you pack, perhaps on Data.fs.old.
OK... but what are your options when it finds an error? :( 491604840 object serialno 0x0346d214c379e7dd does not matchtransaction id 0x0346 d21583aaf791 currently running fsrecover.py on a copy of the Data.fs and hoping for the best...
Damage inside a transaction record may not be detected.
For FileStorage, damage at pickle level and ZODB level (dangling references etc) can be checked with fsrefs.py.
Is this new to 2.6? Zope 2.5.1 doesn't come with any such script. -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's SOLID REPAIRMAN COBRA! (courtesy of isometric.spaceninja.com)
Hello Yes, I also have my stories. The most recent is since yesterday. I cannot work with IE6 with my application. It works fine in the ZMI, also index_html works. But when I call the first form (linked by index_html), it shows me a HTTP 500 error. IE works on an other machine which is ZSynced. On my machine I have to work with Netscape... Regards Dieter PS: Restoring an old data.fs now, and it works again
-----Ursprüngliche Nachricht----- Von: zope-admin@zope.org [mailto:zope-admin@zope.org]Im Auftrag von beno Gesendet am: Montag, 13. Januar 2003 02:22 An: zope@zope.org Betreff: [Zope] The Goblins of Zope
Hi; Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours.
What inspired this post was something that happened tonight. I was playing around with a very simple script when I kept running into errors: Can't convert an integer into a string; Can't convert a string into an integer; etc. I *knew* there wasn't a problem with my script. After a futile hour of trying to track down the *problem*, I decided to copy a perfectly good and currently functioning script, giving it a different name and passing the variables like I do to its parent. Sure enough, I ran into the same errors! Why?!
My worst nightmare happened a couple of weeks ago. I decided to rebuild my file structure into something more portable. I had placed a number of scripts and old DTML methods in the root. I decided to copy them all into a folder, add the folder to each of my sites, and put the new folder into each site's path. Simple, right? What could possibly go wrong? Well, I forget the exact errors Zope threw at me, but it was all utter nonsense. I finally decided I'd just restart the sucker. I mean, why not? Bad, bad idea. I couldn't restart it. Worse yet, I ended up having to reinstall Zope. When I went to open my old Data.fs file, the goblins had secretly returned it to the exact state it was in at my last backup...two weeks before! How???? How could it have possibly known?!
Also, if you happen to know a good exorsict, please let me know <g> beno
_______________________________________________ Zope maillist - Zope@zope.org http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
Dieter Fischer wrote at 2003-1-13 09:41 +0100:
Yes, I also have my stories. The most recent is since yesterday. I cannot work with IE6 with my application. It works fine in the ZMI, also index_html works. But when I call the first form (linked by index_html), it shows me a HTTP 500 error. IE works on an other machine which is ZSynced. On my machine I have to work with Netscape... Does the Zope logfile (--> "LOGGING.txt") tells you something?
Often IE replaces Zope's detailed error messages with its own unhelpful ones. Disable IE's "smart error messages" and you may get a chance to see, what the "server error" (this means HTTP 500) means precisely. It does not look like a Zope Goblin... Dieter
On Monday 13 January 2003 1:22 am, beno wrote:
When I went to open my old Data.fs file, the goblins had secretly returned it to the exact state it was in at my last backup...two weeks before! How???? How could it have possibly known?!
If you have a reproducable example of a data.fs file that is misread by FileStorage at startup, then I am sure that the FileStorage maintainers will be keen to take a look. Is this recorded on the collector?
Also, if you happen to know a good exorsict, please let me know <g> beno
You may want to look at DirectoryStorage, http://dirstorage.sourceforge.net, which is desiged to simplify storage-related exorcisms. -- Toby Dickenson http://www.geminidataloggers.com/people/tdickenson
At 11:02 AM 1/13/2003 +0000, you wrote:
On Monday 13 January 2003 1:22 am, beno wrote:
When I went to open my old Data.fs file, the goblins had secretly returned it to the exact state it was in at my last backup...two weeks before! How???? How could it have possibly known?!
If you have a reproducable example of a data.fs file that is misread by FileStorage at startup, then I am sure that the FileStorage maintainers will be keen to take a look. Is this recorded on the collector?
I honestly didn't think anyone would want to spend their time looking at my problem, so I just rebuilt from the Data.fs I had. It appears, though, that a Data.fs.tmp file remains untainted since the crash.
Also, if you happen to know a good exorsict, please let me know <g> beno
You may want to look at DirectoryStorage, http://dirstorage.sourceforge.net, which is desiged to simplify storage-related exorcisms.
This looks good! Thanks! beno
beno wrote at 2003-1-12 21:22 -0400:
Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours. Apparently, the goblins mostly attack selected people...
I have seen only few weird Zope behaviour examples: 1. someone reported that a folder went suddenly missing 2. the ZCatalog lost suddenly most of its content 3. crashing Zope 4. a few others, of not so many drastic effects Fortunately, the ZODB is *extremely* resilient to loss (unless you pack it). 1. Analysing the ZODB's Data.fs, I could prove that the parent folder of the missing one was deleted and later recreated with "manage_import" and I could recover the lost folder. the state when the fold 2. The "Symlink" product had unindexed the objects. We knew, "Symlink" was dangerous, just not that it is so dangerous. However, the ZODB let us easily recover. Sometimes, Zope is blamed for bugs in other packages: 3. turned out to be a memory corruption in the Oracle 8.1.6 client library... 4. it did not investigate, yet... Dieter
Just my experience to relate; with Zope 2.6.0 compiled from source on Solaris 8, last night I created a DTML document which "disappeared". Thinking I had not hit 'save', I recreated it and then I saw two DTML docs with the same name in the ZMI! I was able to delete one of them and the other remained. Same object id, no title, no description.. Strange. -Patrick Price Dieter Maurer wrote:
beno wrote at 2003-1-12 21:22 -0400:
Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours. Apparently, the goblins mostly attack selected people...
I have seen only few weird Zope behaviour examples:
1. someone reported that a folder went suddenly missing
2. the ZCatalog lost suddenly most of its content
3. crashing Zope
4. a few others, of not so many drastic effects
Fortunately, the ZODB is *extremely* resilient to loss (unless you pack it).
1. Analysing the ZODB's Data.fs, I could prove that the parent folder of the missing one was deleted and later recreated with "manage_import" and I could recover the lost folder. the state when the fold
2. The "Symlink" product had unindexed the objects. We knew, "Symlink" was dangerous, just not that it is so dangerous.
However, the ZODB let us easily recover.
Sometimes, Zope is blamed for bugs in other packages:
3. turned out to be a memory corruption in the Oracle 8.1.6 client library...
4. it did not investigate, yet...
Dieter
On Monday 13 January 2003 03:22, beno wrote:
Hi; Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours.
[SKIPPED] This is a truth that Zope Goblin (actually just a Goblin) is a poor stuff between a chair and keyboard with broken /dev/hands Unix device... -- Regards, Bogdan \|/ ______ \|/ Hey, you've got Blue Screen of Death, "@'/ ,. \`@" and all your data was tweaked by Microsoft Disk Check! /_| \__/ |_\ \___U__/ Please make backup copies to avoid problems in the future...
At 11:02 AM 1/15/2003 +0200, you wrote:
On Monday 13 January 2003 03:22, beno wrote:
Hi; Anyone else had weird, inexplicable things happen with your Zope? Here's a couple of horror stories of mine. I'd love to hear yours.
[SKIPPED]
This is a truth that Zope Goblin (actually just a Goblin) is a poor stuff between a chair and keyboard with broken /dev/hands Unix device...
Well, there's that goblin, true, but the fact remains that strange things happen occasionally on Zope, and all I can say is if you're not aware of that yet, you one day will be ;) beno
On Wednesday 15 January 2003 13:54, beno wrote:
Well, there's that goblin, true, but the fact remains that strange things happen occasionally on Zope, and all I can say is if you're not aware of that yet, you one day will be ;)
Nobody is perfect. Even your "QUALCOMM Windows Eudora Version 5.1"... :) OTOH, Zope is for *FREE*. It means no pay for a tool. But if you get a J2EE, insert a lot of money there, then you will get not only Goblin, but a true big fat Troll with a big hammer (from a Quake 1 or "Lord of the rings") when you will see weird useless Java tracebacks in few browser screens to scroll. -- Regards, Bogdan Remember the... the... uhh.....
At 03:40 PM 1/15/2003 +0200, you wrote:
On Wednesday 15 January 2003 13:54, beno wrote:
Well, there's that goblin, true, but the fact remains that strange things happen occasionally on Zope, and all I can say is if you're not aware of that yet, you one day will be ;)
Nobody is perfect.
And Zope is one he!! of a good product! beno
On Wed, Jan 15, 2003 at 03:40:08PM +0200, Bo M. Maryniuck wrote:
Zope is for *FREE*. It means no pay for a tool. But if you get a J2EE, insert a lot of money there, then you will get not only Goblin, but a true big fat Troll with a big hammer (from a Quake 1 or "Lord of the rings") when you will see weird useless Java tracebacks in few browser screens to scroll.
This corresponds exactly with our experience of Zope and BEA solutions: Zope + CMF: runs for days and weeks without trouble, restarts rarely needed during development, and it restarts in a few seconds. Cost nothing to deploy on several servers. Not free from "goblins" BEA + Portal: runs for a few days at most without trouble, restarts are constantly needed during development, and it restarts in minutes. Cost somewhere in the low hundreds of thousands US dollars to deploy on several servers. Not free from "goblins". OTOH, we're still doing a lot with BEA ... got a lot more Java / JSP knowledge here in house than python / zope knowledge, and they've built some pretty cool apps with it, and I'm the only one who thinks we should use python for everything. :) But we are seriously considering ditching Portal and BEA and moving to something like Tomcat. -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's EX- PEASANT -RECEIVER! (courtesy of isometric.spaceninja.com)
On Wednesday 15 January 2003 20:52, Paul Winkler wrote:
But we are seriously considering ditching Portal and BEA and moving to something like Tomcat.
Nnnnoooo! Tomcat / Apache is a horrible nightmare and the development will cost you much-much-much more that Zope / Python. I don't think Micro$oft are stupid company, who start developing it's dotNet. IOW, reading thin manual of Python, and a little bit bigger manual about Zope is really EASIER than everybody should learn 800 pages of book only of Java language *basis*, then go to the market and buy dozen of fat books to know how to write simple app "Hello world" as XML-RPC webservice and finally write once and DEBUG everywhere... <QUESTION> Does anybody heard about something like .Net and Zope/Python? </QUESTION> -- Regards, Bogdan I am a computer. I am dumber than any human and smarter than any administrator.
<QUESTION> Does anybody heard about something like .Net and Zope/Python? </QUESTION> I'm messing with it right now. Have few .NET goblins already. For example - the most important event (BeforeNavigate2) being broken in C# , which kills the possibility of robust 'permanent' link between .NET part and Zope admin UI. There is enough of architectural problems on both sides. However, something nice *is* possible. For example, I already have the .NET part being able to perform the atomic admin things transparently (deleting/adding folders and other stuff) With plain HTTP, not with XML-RPC. Rgds.Paul. PS. This is in context of Desktop Zope : http://www.pault.com/pault/zope/dz.html PPS. If there would be enough interest, I have no problem opening the source code of .NET part. Should get the first application complete in two weeks or something. Write me if interested.
On Thursday 16 January 2003 14:06, Paul T \(spk\) wrote:
I'm messing with it right now. Have few .NET goblins already. :-) Have you some links?
-- Regards, Bogdan If you are smart enough to know that you're not smart enough to be an Engineer, then you're in Business.
On Thursday 16 January 2003 14:06, Paul T \(spk\) wrote:
I'm messing with it right now. Have few .NET goblins already.
:-) Have you some links?
As to broken BeforeNavigate2 : http://groups.google.com/groups?q=beforenavigate2&hl=en&lr=&ie=UTF-8&oe=UTF- 8&selm=e3Ht6aKACHA.2216%40tkmsftngp02&rnum=1 There are some 'workarounds' on the web, but those are not actually workarounds ( 'string headers' of BeforeNavigate is *not* 'ref Array headers' ) Of course, ignorant Windows developer would not consider this to be a goblin of .NET because there are plenty of articles, which : 1. show how to do that in C++ ( it works in C++. It is broken only in C# ). 2. explaining how to 'fix it' in C# ( only the problem is that the fix is not the fix, but 'almost a fix' ) Rgds.Paul. PS. Another .NET goblin is related to POST. Try using WebControl uploadForm on Zope's administrative UI - you may see some funny things. Sorry, have no link for that yet, but may send you the tiny snippet of the source code eventually. Which reminds me of the fact that POST was broken in Tomcat/Linux for ages and nobody did fixed it (Tomcat was open source). PPS. I also found that the Windows developers have special attitude, regarding subtle bugreports. (Windows has many sources of subtle bugs, for example, some versions of IE can process some client-side XSLT, some can not e t.c..) So, when presented with the bugreport, the first reaction of novice Windows developer usually is "it works on my computer, there is something wrong with your hands". I think it is funny.
beno wrote:
Well, there's that goblin, true, but the fact remains that strange things happen occasionally on Zope, and all I can say is if you're not aware of that yet, you one day will be ;)
I've never had anything I'd class as 'strange things' happen with Zope in 4-odd years of use... cheers, Chris
I agree with Chris. We've been using Zope for just over a year and we haven't had any strangeness that wasn't without good reason. Almost all of the problems that we have had have been because of programming errors on our part, configuration errors, or third-party product problems. We've had production sites up for many months now with no trouble. One of the nice things about Zope is that if there is some weird going on, you have the source code to fix it. Try that with commercial J2EE platforms. Kevin Chris Withers wrote:
beno wrote:
Well, there's that goblin, true, but the fact remains that strange things happen occasionally on Zope, and all I can say is if you're not aware of that yet, you one day will be ;)
I've never had anything I'd class as 'strange things' happen with Zope in 4-odd years of use...
cheers,
Chris
_______________________________________________ Zope maillist - Zope@zope.org http://lists.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope-dev )
At 12:41 PM 1/15/2003 -0500, you wrote:
I agree with Chris. We've been using Zope for just over a year and we haven't had any strangeness that wasn't without good reason. Almost all of the problems that we have had have been because of programming errors on our part, configuration errors, or third-party product problems. We've had production sites up for many months now with no trouble.
One of the nice things about Zope is that if there is some weird going on, you have the source code to fix it. Try that with commercial J2EE platforms.
You guys are all much better programmers than I. I certainly don't want to give anyone the impression that I'm complaining. Nothing's perfect. Zope is a wonderful product. I'm the weakest link in the lot. But what happened just looked so weird...and I've had similar things happen before. I was just curious if anyone else had, too. No need to rush to the defence of an excellent but imperfect product :)) beno
On Wed, Jan 15, 2003 at 05:28:48PM +0000, Chris Withers wrote:
I've never had anything I'd class as 'strange things' happen with Zope in 4-odd years of use...
I have, but it's only an "odd thing" until you figure out what triggered it. There are some things I have not yet figured out. E.g. all the portal type icons are now missing from CMF but ONLY if I view it in IE 5.5, every other browser / platform is OK - and IE only started doing this last week. Is this Zope's "odd thing" or IE's? Hmmmm. :) -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's MARKSMAN PICNINC ! (courtesy of isometric.spaceninja.com)
On Wednesday 15 January 2003 20:54, Paul Winkler wrote:
E.g. all the portal type icons are now missing from CMF but ONLY if I view it in IE 5.5 This is a common problem: PNG's are not shown SOMETIMES in M$IE 5.5 and real fat Goblin here is a Microsoft, as it actually is.
-- Regards, Bogdan MSDOS didn't get as bad as it is overnight -- it took over ten years of careful development. -- dmeggins@aix1.uottawa.ca
On Thu, Jan 16, 2003 at 01:31:22PM +0200, Bo M. Maryniuck wrote:
On Wednesday 15 January 2003 20:54, Paul Winkler wrote:
E.g. all the portal type icons are now missing from CMF but ONLY if I view it in IE 5.5 This is a common problem: PNG's are not shown SOMETIMES in M$IE 5.5 and real fat Goblin here is a Microsoft, as it actually is.
I am not surprised at all. :-| The weird thing is that it used to consistently work, and now it consistently does not work. I wonder what Windows changed without telling me. :-( I'd upgrade to 6.0 but I need to keep 5.5 around to check compatibility of my html... meantime i try to do all my zope development in mozilla. :) -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's MAILMAN PICKLE! (courtesy of isometric.spaceninja.com)
beno wrote alot of things But, error reports er worthless if they do not describe how the error may be reproduced. There is always a logical reason. We are doing computer science not voodoo, so please let the Goblins stay in the FairyWorld. :) \Oliver -- Direkt?r / Managing Director Oliver Marx TEKK Lyngbyvej 20 DK-2100 K?benhavn ? Main: +45 39 15 80 60 Direct: +45 39 15 80 62 http://www.tekk.dk
At 02:49 PM 1/15/2003 +0100, you wrote:
beno wrote alot of things
But, error reports er worthless if they do not describe how the error may be reproduced. There is always a logical reason.
Of course, and if I could have reverse engineered what happened, I would have shared it. But I'm not the sharpest tool in the shed ;) I just know something went wrong.
We are doing computer science not voodoo
Even voodoo is science that we simply don't understand ;) beno
participants (13)
-
beno -
Bo M. Maryniuck -
Charlie Reiman -
Chris Withers -
Dieter Fischer -
Dieter Maurer -
Kevin Carlson -
Oliver Bleutgen -
Oliver Marx -
Patrick Price -
Paul T (spk) -
Paul Winkler -
Toby Dickenson