- Zope-Dev - Zope lists

RE: [Zope-dev] Cache-bug in handling of files
by Brian Lloyd 11 Aug '00

11 Aug '00

> I have fileobjects where I store pdf-files. If I want to change the > pdf-file I use the method "manage_upload". The problem is > that if I try > to view it it always choose the one that is in the cache. It displays > the new one if I open it in a new window or if I save it to disk. The > filesize and everything else is for the new file. That sounds very suspicious - what are your IE cache settings? > This points to a bug report and after serching for it I found it: > > [754] Zope Bug: Uploaded Images Don't Appear Immediately > Category: General Last Updated: Oct 29, 1999 7:05 am > Submitted On: Oct 12, 1999 3:58 am Status: Ongoing > Reported Version: 2.0.1 Expected/Fixed Version: > > The strange thing is that this is still Ongoing after almost a year > since it was submitted! This bug report was related to the fact that the user was accessing the file through a Squid cache and that at the time (Zope 2.0.1) there was a legitimate bug in the way that "If-Modified-Since" headers were handled. Squid was actually returning the file to him. It should have been marked completed quite a while ago (I just did that). I am using IE 5 (5.00.2314.1003), cache settings set to: "Check for newer versions of stored pages: x Every visit to the page ...and I cannot reproduce the problem that you are describing :( Are these your cache settings? Are you accessing the server through any kind of proxy? Thanks Brian Lloyd brian(a)digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com

2 1

Cache-bug in handling of files
by Peter Arvidsson 11 Aug '00

11 Aug '00

I started this thread on the regular zope mailinglist but now I found a similar posting to my problem in the archives. I have fileobjects where I store pdf-files. If I want to change the pdf-file I use the method "manage_upload". The problem is that if I try to view it it always choose the one that is in the cache. It displays the new one if I open it in a new window or if I save it to disk. The filesize and everything else is for the new file. The thread I found was this: http://zope.nipltd.com/public/lists/zope-archive.nsf/Main/AED89A93819422558… This points to a bug report and after serching for it I found it: [754] Zope Bug: Uploaded Images Don't Appear Immediately Category: General Last Updated: Oct 29, 1999 7:05 am Submitted On: Oct 12, 1999 3:58 am Status: Ongoing Reported Version: 2.0.1 Expected/Fixed Version: The strange thing is that this is still Ongoing after almost a year since it was submitted! There was a fix for the problem in the mailinglist thread but 1. I dont know if Image.py is used for all files and 2. I looked in Image.py and didnt fing the code to be similar to the one in the thread. So because I am a newbie in python I want to ask if any1 could take a look at this for me and hopefully make a patch or at least get as much information I need to chunk it in the collector. Peter

1 1

What should be done with Nulls in Database result sets?
by Jay, Dylan 11 Aug '00

11 Aug '00

I have been having all sorts of trouble upgrading from ZODBCDA to ZmxODBC. The ZODBCDA users RDB.DatabaseResults and ZmxODBC (modified from ZOracleDA) uses Results.Results. These two classes seem to do exactly the same purpose and so I suppose one shouldn't exist. The important point is that handle NULL values very differently. Results let them pass through as None values whereas DatabaseResults change them to the special value MissingValue. MissingValue is nice becuase its string representation is an empty string which is nice on html forms etc. Why doesn't Results do this too? What is the official word on what a DA should be returning for NULLs? ---- Dylan Jay Tel: +61 2 9886-8961 Avaya Communication FAX: +61 2 9352 9224 Level 5, 15 Talavera Road Mobile:+61 409 606 171 Nth Ryde, NSW, 2113 ICQ: 520341 AUSTRALIA

1 0

RE: [Zope-dev] OFS.objectManager checking object Ids
by Jay, Dylan 11 Aug '00

11 Aug '00

> -----Original Message----- > From: Jim Fulton [mailto:jim@digicool.com] > Sent: Thursday, August 10, 2000 9:51 PM > To: David C. Kankiewicz > Cc: Jay, Dylan; 'zope-dev(a)zope.org' > Subject: Re: [Zope-dev] OFS.objectManager checking object Ids > > "David C. Kankiewicz" wrote: > > > > "Jay, Dylan" wrote: > > > > > > As I said, when the url is quoted there is no problem. > > > > > > > http://azonia.auslabs.lucent.com/Auslabs/AuslabsFAQ/Where%20is%20X%3f > > > > > > causes no problems. The above object is sitting happily > inside Zope with a > > > question mark at the end. I had to let the bad_id check > let it pass but is > > > there any reason why it shouldn't? > > > > bad_id=ts_regex.compile('[^a-zA-Z0-9-_~\,\. ]').search #TS > > > > The .search was .match > > Changing it back restores the behavior, who changed it and > why is the > > question? > > I don't know who changed it, but the why is obvious. > 'match' was clearly a bug, since it only found an > invalid character if it was at the beginning of an id. > > I think that it's a bad idea to allow '?'s in ids > and am sorry if it was allowed. In general, I don't > like to see characters in ids that need to be quoted. > I'm not happy that ' ' was added, although > I understand why. It may seem like a wise idea not to allow such chars into an id but I'd to explore this issue a little more as I think it is not quite as trivial as it sounds. The proposition I'm putting forward is that, if Zope copes with it why not let it. In dtml your already have to quote things like sequence-item becuase its not valid python but people cope with that. What we are talking about is have a sensible match between what an item is and what its called. I know that that is what the title property is for however look at the example that lead me to this problem in the first place. I want an FAQ that people in my department can all add too. None of the existing FAQ's are good enough. They arn't flexible or customizable enough. So I decided to use ZWiki which is flexible and customizable. But ZWiki needs WikiNames which become the object ids. A FAQ contains questions. Questions don't make good wiki names but structured text lets me make anything into a WikiName eg [How do I turn a ZWiki into a collaborative FAQ?] So I create a dtml method that appends this text to the end of a wiki page and create the resulting question page (with the question mark in the id). This makes it a reasonably usable FAQ system with all the benifits of a ZWiki. People can move questions around my cutting and pasting. I can let people see who is editing what with a little bit of coding here and there. So without the question directly in the id I'm left with the "how to create a unique id" and then I have reference it within pages using the title anyway becuase I don't want people seeing the random ID. Why should I have to go to so much trouble when I can just call the object by the title anyway and Zope has no problem with it. It makes it easier for me and I'm sure it will make it easier for other uses to. Resorting to random names in urls is ugly and unneccerily complicated. If Zope fundumentally can't cope with strange ids then say so but I've seen no evidence of this. Anyway, thats my 2c worth.

1 0

Re: "permission mapping error" when importing a ZClass-based product.
by Anthony Baxter 10 Aug '00

10 Aug '00

Ok, I got a quick workaround by making a new Product, and creating the permissions it needed inside there. I _suspect_ the problem was caused by nested ZClasses. Uuuugly. Anthony >>> Anthony Baxter wrote > > I'm trying to export/import a ZClass based product from one Zope2.2 > system to another, and I'm getting the following: > > Error Type: Permission mapping error > Error Value: Attempted to map a permission to a permission, Edit ekit Partn er, > that is not valid. This should never happen. (Waaa). > > Well, it has happened, and I'd like it to go away. Any ideas what causes > this? > > Anthony > -- > Anthony Baxter <anthony(a)interlink.com.au> > It's never too late to have a happy childhood. > -- Anthony Baxter <anthony(a)interlink.com.au> It's never too late to have a happy childhood.

1 0

"permission mapping error" when importing a ZClass-based product.
by Anthony Baxter 10 Aug '00

10 Aug '00

I'm trying to export/import a ZClass based product from one Zope2.2 system to another, and I'm getting the following: Error Type: Permission mapping error Error Value: Attempted to map a permission to a permission, Edit ekit Partner, that is not valid. This should never happen. (Waaa). Well, it has happened, and I'd like it to go away. Any ideas what causes this? Anthony -- Anthony Baxter <anthony(a)interlink.com.au> It's never too late to have a happy childhood.

1 0

Re: [Zope-dev] OFS.objectManager checking object Ids
by Dieter Maurer 10 Aug '00

10 Aug '00

David C. Kankiewicz writes: > "Jay, Dylan" wrote: > > > > As I said, when the url is quoted there is no problem. > > > > http://azonia.auslabs.lucent.com/Auslabs/AuslabsFAQ/Where%20is%20X%3f > > > > causes no problems. The above object is sitting happily inside Zope with a > > question mark at the end. I had to let the bad_id check let it pass but is > > there any reason why it shouldn't? > > > bad_id=ts_regex.compile('[^a-zA-Z0-9-_~\,\. ]').search #TS > > The .search was .match > Changing it back restores the behavior, who changed it and why is the > question? > > > I can't find any reason why, atleast not right now. > Maybe I need more sleep :), didn't remember the URL quoting stuff... '?' is bad because it requires you to quote all names you are going to use as parts of URL's. Non-ASCII characters, too, are rightfully forbidden: a collegue used a 'ä' in an id. Worked fine for browsers on Windows and Linux but broke for browsers on a Mac's. Dieter

1 0

Re: [Zope-dev] Weird Data.fs corruption... zope-dev@zope.org
by Dieter Maurer 10 Aug '00

10 Aug '00

Martijn Pieters writes: > Ai, NFS! There is a known no-no about using NFS for Data.fs storage, it leads > to data corruptions. Zope and NFS do not mix. Huch, why that? I use Data.fs (occasionally) over NFS and did not yet have had problems. And appending to a file should work over NFS. Locking may be a bit more problematic, but this is not an issue provided I ensure that only on Zope process runs at a time. Dieter

3 2

Re: [Zope] ZPatterns Question
by Steve Alexander 10 Aug '00

10 Aug '00

I've moved your question to zope-dev, as that's where ZPatterns questions probably belong. Roche wrote: > I created a ZClass called "Product" with Dataskin as baseclass. > > I also added a specialist called ProductManager containing a default rack. > Under storage for the rack I selected the newly created Product ZClass. > > My product ZClass needs to query both a RDBMS and the ZODB for > attributevalues. At this point I get lost. > I guess i should add an attributeprovider plug-in??? The easiest way is to use a couple of SheetProviders. Use a PersistentSheetProvider for your ZODB values. You'll have to write your own one for SQL values. You will need to create the appropriate propertysheets in your ZClass class definition in your product. > What should go into the fromexpr and attrsexprs properties of the > attributeprovider? Should one add an attributeprovider for each attribute > of the product? See the IRC chat log: http://www.zope.org/Documentation/Chats/zpatt_chat_log An attribute provider can provide multiple attributes. You can have multiple attribute providers if you like. -- Steve Alexander Software Engineer Cat-Box limited http://www.cat-box.net

5 8

RE: [Zope-dev] RE: [Zope] ZPatterns Question
by RC Compaan 10 Aug '00

10 Aug '00

> At 01:57 PM 8/8/00 +0200, RC Compaan wrote: > >I've added a propertysheet called "properties" to my ZClass and i notice > >there is Persistent Sheetprovider under the default rack already. The > >Sheetprovider has properties Sheet_Names and Sheet_Namespaces. I guess > >Sheet_Names should refer to the sheetname i created for my > ZClass??? but how > >do Sheet_Namespaces come into play. > > Actually, neither relates. Property sheets created on ZClasses > always have > their data stored in attributes of the object itself, so if you want to > control those property sheets you would not use a sheet provider at all. > Sheet providers are only used to provide property sheets whose data is > stored external to the object. > Oh ok, this divide clears it up a bit. I somehow imagined that an object's properties are always stored inside the object and that Providers are the managers of those properties. > The Specialist does not talk to the sheet provider. The Specialist asks > its racks for an object. The Rack which "retrieves" the object tells the > object that it belongs to the rack and should ask the rack for anything it > needs. When the object needs a property sheet or attribute or > whatever, it > asks the rack to give it a list of relevant providers for that > attribute/sheet/whatever. It then walks the list asking those > providers if > they can perform the function it desires. If none of the providers are > successful, it performs a default behavior (such as raising an > AttributeError to indicate the attribute does not exist). This general > pattern is followed for almost anything that can be delegated from a skin > to a provider. I considered the scenario where attributes are stored external to the object according to the filing clerk analogy on the wiki: The Specialist (filing clerk) asks its racks (filing cabinet) for an object (dental records). The rack (filing cabinet) tells the object (the dental records) that it belongs to the rack AND that the object should ask the rack for anything it needs. Why would the object "need" anything. Is it not the specialist with the "need"? If object data is stored external to the dataskin object does the dataskin only implement behaviour? I find it difficult to understand because I associate Dataskin with Storage. If object data is stored externally is it stored in Persistent Property Sheets? How does one define properties for Persistent Property Sheets? Roché

2 1