February 1999 - Zope

Location of sequence-previous, sequence-next
by Shaw, Howard 17 Feb '99

17 Feb '99

I have not looked at the internal code, so I don't know what is involved, but I would very much prefer it if the 'sequence-previous' data and the 'sequence-next' data were each available at every phase of the  tag. I would like to do:  <tr> <td>Blah </td> <td>Blah </td> <td>Blah </td> <td>Blah </td> </tr>  Previous href Next href   where basically both the previous pointer and the next pointer show up at the end of the table. At the moment I am using this kludge:   <tr> <td></td> <td></td> <td><a href="mailto:"></td> <td><a href="mailto:"></td> </tr>    <tr> <td></td> <td></td> <td><a href="mailto:"></td> <td><a href="mailto:"></td> </tr>   <tr><td colspan=4><table width=100%><tr>    <td><center><a href="query_start=">Previous</a></center></td>  <td><center>Previous</center></td>     <td><center><a href="query_start=">Next</a></center></td>  <td><center>Next</center></td>    I don't know what the overhead is for performing that  twice, but I don't like it. It makes the code look kludgey and is a pain to work with. This code would be much cleaner and more understandable if the 'previous-sequence' stuck around until 'sequence-end'. Or for that matter, if there was a way, within an if block, to set variables in the surrounding namespace as opposed to an internal #with grouping. Then I could do like this:  <tr> <td>Blah </td> <td>Blah </td> <td>Blah </td> <td>Blah </td> </tr>  Set 'my_previous_blah' equal to 'previous_blah'   Previous href (using my_sequence-prev) Next href (using sequence-next)   which would still be a kludge, and yet still be easier to understand.

1 0

REQUEST.unset
by Service Informatique CHSR 17 Feb '99

17 Feb '99

Hello, It is useful to have the opposite of REQUEST.set, that is REQUEST.unset to delete an application variable. REQUEST.unset('name') is intended to act the same as: del REQUEST.other['name'] Included is a patch to be applied to lib/python/ZPublisher/Request.py What's your opinion about this guys? Should I submit this to the collector? Regards, Jephte CLAIN Service Informatique CHSR

1 0

something like , but outside  tag
by Service Informatique CHSR 17 Feb '99

17 Feb '99

Hello, I wonder if something behaving like  but outside the  tag exist? Regards, Jephte CLAIN Service Informatique CHSR

1 0

Major error in my MailHost patch
by Martijn Pieters 17 Feb '99

17 Feb '99

Hello folks, I was majorly tired last night, and goofed up big time with the most recent patch. It contains a major bug, and your MailHost.py won't even compile. So, don't use right now! Am busy correcting it. -- M.J. Pieters, Web Developer | ATMM http://www.atmm.nl | Tel: +31-35-6254545 Fax: +31-35-6254555 | mailto:mj@atmm.nl http://www.atmm.nl/~mj | PGP: http://wwwkeys.nl.pgp.net:11371/pks/lookup?op=get&search=0xA8A32149 ------------------------------------------

1 0

Re: [Zope] ZOPE ODBC on Linux
by Phil Harris 17 Feb '99

17 Feb '99

What an interesting thought, I'm going to give this a go 8^) Phil ----- Original Message ----- From: David Ascher <da(a)ski.org> To: Hannu Krosing <hannu(a)trust.ee> Cc: Randall Marbach <remarbach(a)netscape.net>; <Zope(a)zope.org> Sent: Tuesday, February 16, 1999 10:31 PM Subject: Re: [Zope] ZOPE ODBC on Linux >On Wed, 17 Feb 1999, Hannu Krosing wrote: > >> Randall Marbach wrote: >> > >> > Is it possible for a Zope 1.9 installation running on a linux platform (redhat >> > 5.2) to access a MS ACCESS database located on a different NT server via the >> > ODBC support built into Zope? >> >> As Access is not a client-server database, you are unable to use it >> even from another Windows (NT or 9x) machine. > >Uh, you could do something complicated involving e.g. RemoteCall from >Linux to a Win32 box, and a PythonCOM program on Win32 box. > >--david > > >_______________________________________________ >Zope maillist - Zope(a)zope.org >http://www.zope.org/mailman/listinfo/zope >

1 0

Re: [Zope] ZOPE ODBC on Linux
by Phil Harris 17 Feb '99

17 Feb '99

Try http://www.openlinksw.com/iodbc HTH Phil ----- Original Message ----- From: Christopher G. Petrilli <petrilli(a)amber.org> To: <Zope(a)zope.org> Cc: <remarbach(a)netscape.net> Sent: Tuesday, February 16, 1999 10:02 PM Subject: Re: [Zope] ZOPE ODBC on Linux >On Tue, Feb 16, 1999 at 01:38:27PM -0800, Randall Marbach wrote: >> Is it possible for a Zope 1.9 installation running on a linux platform (redhat >> 5.2) to access a MS ACCESS database located on a different NT server via the >> ODBC support built into Zope? > >Well, this would require an ODBC driver under Linux, which obviously >Microsoft doens't provide. I don't currently know of any ODBC driver >that is freely available that runs on a UNIX variant. > >> If not, is there some other way to do this? > >You might check out... >http://starship.skyport.net/crew/lemburg/mxODBC.html > >It might have some more information. (un)Fortunately I talk native to >all my databases, but ODBC is "native" for MS products. > >Chris >-- >| Christopher Petrilli >| petrilli(a)amber.org > >_______________________________________________ >Zope maillist - Zope(a)zope.org >http://www.zope.org/mailman/listinfo/zope >

1 0

Feature Request Withdrawal(RE Non-English User Names)
by LEE, Kwan Soo 17 Feb '99

17 Feb '99

*I have posted almost same message using www.egroups.com/list/Zope from an internet cafe just before the New Year's Day Holiday Season on Lunar Calendar began. But interestingly, that message does not seem get to the list and I got an mail from support(a)digicool.com on the last day of holidays on my OFFICE!(I just dropped in to see everything going well). Sorry & poor those at DC. I am withdrawing my former Feature Request(numbered 22 or 23), since what I asked(see below) is not your fault but those at Netscape. Waiting for DC's reply, I have hacked around the User.py(it took several days since I am an awful newbie at python & programing) to see what's goinng on myself. Now I can definately say, Lynx & MS IE 4.x(English version & Korean version) do not cause any problem with Non-English User Name, but Netscape Navigator DOES no matter what language version and on what os. It simply delivers only a part of the whole auth string. I have verified this on Win95, Win98 & Linux using various versions of above mentioned browsers at several places which uses different methods to be connected to the net(that's why I posted the undelivered mail from an internet cafe not from my office). And posted a bug report to Netscape. Sincere apology for the nuisance that my hassle caused. : - ( LEE Kwan Soo. PS, I suppose that Netscape Navigators for other languages have the same problem. If it happens to be a concern for you, please 'bug report' to Netscape to press them quicker remedy. support(a)digicool.com wrote: > LEE Kwan Soo reported: > > > I can add a new entry whose id has non-English characters(in my case Korea) into UserFolder and UserDB successfully. > > But accessing management screen as that user raises Bin.ascii error. > > > > "decodestring" imported from base64.py module and used in User.py line 388 causes that error. > > > > Don't you think you shoud not allow non-English username if this error is not enevitable? > > > > Or better I'd like to be able to use Non-English usernames. > > > > ps. If those at Digicool found this problem has low priority, I guess, I might able to produce a quick & ugly hack to get around this problem when you tell me how & with what line 388 of User.py get username and password. > > We'll investigate this.

1 0

Tree Tag Options
by Martijn Pieters 17 Feb '99

17 Feb '99

In addition to my previous email, here are two more attributes: header="object" If set, this property is assumed to be an object with a DTML rendering. It is displayed before the current object. The header DTML method is handed a special standard_html_header and standard_html_footer. footer="object" If set, this property is assumed to be an object with a DTML rendering. It is displayed after the current object. The header DTML method is handed a special standard_html_header and standard_html_footer.

1 0

Tree Tag Options
by Martijn Pieters 17 Feb '99

17 Feb '99

At Tue, 02 Feb 1999 13:01:32 -0800 Amos Latteier wrote a nice explanation of the branches attribute of the tree tag. But when I played around with the source of the tree tag (TreeTag.py), I discovered more toys. This is has become an unofficial documentation of the #tree tag. Excuse the typos, for it has become very late. Have fun. Attributes of #tree assume_children=[0|1] default 0 If set to a nonzero value, all items are assumed to have subitems, and will therefore allways have a + sign in front of them when not yet expanded. Only when a user expands an item it is checked for subitems. This could be a good option when the retrieval of children is a slow process. branches="methodname" default="tpValues" If an object has defined this method, it should return a list of ids that are it's subobjects. The default, tpValues, is defined on the Items class (from which just about every objetc inherits), and Folders and Database Adapters define their own versions. If you use objectValues as method, you get the same effect on Folders, but not on Database Adapters. branches_expr="expression" branches_expr works like the expr attribute of tags like #if. You could therefore do:  to get a tree of only Folder and Document subobjects. This is easier than defining your own External Method for branches="", like Amos described. id="method_or_propertyname" default="tpId" Method or property name to determine the ID of the object for tree-state purposes. The Ids saved in the tree state cookie (and the tree-state local var) are realy just integers so the state var is not too big, but unusable for your average Zope developer. If you want to do anything with the state, you might want to define this as id="id", for a more readable form. Just do  and see for yourself. No default Zope object currently defines a tpId method or property. When no tpId method or property is defined, an internally generated integer is used (pickle id?). leaves="property" If set, this property is assumed to be a DTML Method within objects that have no sub-objects. It is displayed as part of the tree, no additional code is needed in your DTML. The leave DTML method is handed a special standard_html_header and standard_html_footer. nowrap=[0|1] default 0 If set to a nonzero value, the tree table will be rendered with the NOWRAP attribute, which means the browser won't wrap the item labels when spacially challenged. url="method_or_propertyname" default="tpURL" Method or property to be called on object to determine the relative URL of the object. The default, tpURL, is defined by the Item class, and the Gladfly Database Adapter defines it's own version, so it can show you it's tables and columns in the management screen. If you want to override the URL translation, define an External Method, like Amos described, and use that method with url="methodname". single=[0|1] default 0 If set to a nonzero value, only one branch on a level can be expanded. Any other expanded branches will collapse. Seriously cool. skip_unauthorized=[0|1] default 0 If set to a nonzero value, no ValidationErrors will be raised trying to read objects to which the user has no access. Variables in namespace set by tree tag (for every object being rendered): tree-root-url Relative URL of the DTML Document or Method this tag is in. tree-colspan The number of levels deep the tree is being rendered. tree-state The current state of the tree, a list of ids and sublists. tree-item-url The URL of the current object being rendered, relative to the document the #tree tag is in. This is generated using the url="methodname" method. tree-level Level of this object in the tree, starting at 0. tree-item-expanded Flag indicating wether this item is expanded or not. Variables in namespace that influence the tree tag: expand-all All branches expanded if this name is defined and nonzero (like in a cookie, or in the REQUEST object. Try http://URL/to/object/with/tree/tag?expand-all=1) collapse-all All branches collapsed if present and nonzero. tree-s This is the cookie and query name that contains the state of the tree. This is a zlib compressed, base64 encoded nested list of ID's. Have fun making one yourself. I think you better start with the TreeTag.py code. tree-e List of ID's to expand. See tree-s on the format. tree-c List of ID's to collapse. Again, see tree-s. Are we going to collect this kind of data somewhere? Like in a tag reference? Is there a protocol to define your own DTML tags?

1 0

RE: [Zope] ZOPE ODBC on Linux
by Paul Everitt 16 Feb '99

16 Feb '99

> > Randall Marbach wrote: > > > > > > Is it possible for a Zope 1.9 installation running on a > linux platform (redhat > > > 5.2) to access a MS ACCESS database located on a > different NT server via the > > > ODBC support built into Zope? > > > > As Access is not a client-server database, you are unable to use it > > even from another Windows (NT or 9x) machine. > > Uh, you could do something complicated involving e.g. RemoteCall from > Linux to a Win32 box, and a PythonCOM program on Win32 box. Hate to talk about something we aren't publicly releasing, but...we've been doing this for about two years now. We had a Principia Product that called a Network Client that allowed a local object to be a proxy for a remote "searchable object" (read: SQL Method). The two processes communicated via HTTP. The initial request was marshalled, rotor'd, then base64'd and packed in the body of the HTTP request to the second server which ran the query and packaged up the results in the same way. Tracebacks on the server were re-raised on the client process. We later added caching. Thus for a long time the digicool.com site, running on Linux, authenticated users against data stored in Lotus Notes on NT. The response time was pretty spiffy. This project came from the original customer project that created our relational database integration back at the end of 1996. We federated queries for the Department of Defense across a Sybase and Oracle databases running on different machines in different time zones. We continue to use it in our consulting gigs (in fact, a customer that is onsite here today is getting it). I imagine some enterprising soul could craft up a cool solution using XML-RPC or the like... --Paul Paul Everitt Digital Creations paul(a)digicool.com 540.371.6909

1 0