robustness in management interface.
Hi, The zope management interface has some robustness problems: whenever you call manage_workspace (the normal way of managing a folder through the HTML Zope management interface) on a folder X, and some object y in that folder gives an error (fi, it has no title attribute, ) the whole folder becomes unmanageble. The only thing you can do at that point is to manually delete the problematic object by typing: http://..../X/manage_delObjects?ids=y It seems to me that an object in a folder should not interfere with the management of the folder (at least, you should be able to throw the object out of the folder) probably, a try/except or <dtml-try> in the right place fix this. have fun, Sloot.
Romain Slootmaekers wrote:
Hi,
The zope management interface has some robustness problems:
whenever you call manage_workspace (the normal way of managing a folder through the HTML Zope management interface) on a folder X, and some object y in that folder gives an error (fi, it has no title attribute, ) the whole folder becomes unmanageble. The only thing you can do at that point is to manually delete the problematic object by typing:
http://..../X/manage_delObjects?ids=y
It seems to me that an object in a folder should not interfere with the management of the folder (at least, you should be able to throw the object out of the folder)
probably, a try/except or <dtml-try> in the right place fix this.
Or, you could assure that your objects all have titles (at least ="") if you want them managed through the ZMI. I have not seen this as a *requirement*, but every example of a zope object I have seen uses self.title=aString in __init__(). It is also important to have titles for most cataloging. Do you catalog your site? Is not title a good thing for searching? Might you catalog in the future? Perhaps something for a BestPractices document(?) or wiki(?): "ZMI-manageable objects have a title attribute. This is a string." For a bit of context on the above, I put together a product that has title as a function (=[:30] of some content) some time ago. I have been led to understand that this was a bad idea because it breaks some cataloging. Your idea of <dtml-try> does have merit IMHO. Should the ZMI really assume existence of anything other than "id" for objects? -- Jim Washington
Jim Washington wrote:
Romain Slootmaekers wrote:
Hi,
The zope management interface has some robustness problems:
whenever you call manage_workspace (the normal way of managing a folder through the HTML Zope management interface) on a folder X, and some object y in that folder gives an error (fi, it has no title attribute, ) the whole folder becomes unmanageble. The only thing you can do at that point is to manually delete the problematic object by typing:
http://..../X/manage_delObjects?ids=y
It seems to me that an object in a folder should not interfere with the management of the folder (at least, you should be able to throw the object out of the folder)
probably, a try/except or <dtml-try> in the right place fix this.
Or, you could assure that your objects all have titles (at least ="") if you want them managed through the ZMI. I have not seen this as a *requirement*, but every example of a zope object I have seen uses self.title=aString in __init__(). It is also important to have titles for most cataloging. Do you catalog your site? Is not title a good thing for searching? Might you catalog in the future?
maybe I didn't express myself as clearly as I could have: the title attribute was an example (fi=for instance) of what can go wrong.
Perhaps something for a BestPractices document(?) or wiki(?):
"ZMI-manageable objects have a title attribute. This is a string."
For a bit of context on the above, I put together a product that has title as a function (=[:30] of some content) some time ago. I have been led to understand that this was a bad idea because it breaks some cataloging.
yep.your catalog can be inconsistent, unless the changing of the attribute will call a recatalog orso.
Your idea of <dtml-try> does have merit IMHO. Should the ZMI really assume existence of anything other than "id" for objects?
Sloot.
On February 7, Jim Washington wrote:
Or, you could assure that your objects all have titles (at least ="") if you want them managed through the ZMI.
It seems you have missed the author's point: the ZMI should be robust enough to deal with this sort of thing, otherwise it is not very useful as a Management Interface. Also, one doesn't always have full control over all the objects in a folder: for example, what if you import an object that then breaks the ZMI?
Your idea of <dtml-try> does have merit IMHO. Should the ZMI really assume existence of anything other than "id" for objects?
Definitely not. a. -- Adrian van den Dries adriand@flow.com.au Development team www.dev.flow.com.au FLOW Communications Pty. Ltd. www.flow.com.au
On Sunday 09 February 2003 11:16 am, Adrian van den Dries wrote:
Your idea of <dtml-try> does have merit IMHO. Should the ZMI really assume existence of anything other than "id" for objects?
Definitely not.
Its worse than that ;-) The management interface currently assumes that objects do not have certain attributes. It is possible to break the management interface using the 'properties' tab, by adding attributes with carefuly chosen names. It will be hard to fix all of this before zope 3. -- Toby Dickenson http://www.geminidataloggers.com/people/tdickenson
participants (4)
-
Adrian van den Dries -
Jim Washington -
Romain Slootmaekers -
Toby Dickenson