December 1998 - Zope

Re: [Zope] - Setting __bobo_realm__
by Jim Fulton 08 Dec '98

08 Dec '98

"Christopher G. Petrilli" wrote: > > On Tue, Dec 08, 1998 at 07:29:32PM +0000, Jim Fulton wrote: > > "Christopher G. Petrilli" wrote: > > > [ I make Jim's life more difficult about Realms in HTTP ] > > > > > > Ah, what I'm interested in is doing it INSIDE the GUI, on a folder > > > level, not on a "Module" level, or a global level. For example, if I > > > access a folder with it's own Userfolder, and authorize on that, then > > > want to go to the MAIN /manage screen, I have to reauthorize, which is > > > fine, since they both use "Main" as the realm. BUT when I go BACK to > > > the sub folder (above), I'm still running as the "superuser" or > > > whatever. > > > > And that's not what you want? :) In general, we prefer that > > people who are authorized above should be authorized below. > > Well, yes but... :-) Don't you hate that? No, I very much prefer it. If I have an object A/B, I expect the manager of A to be able to manage B and to be able to create managers for B without giving up it's own access to B. > > > Do you *really* want subfolders to have different realms, > > do you just want to be able to able to "log out"? > > Realms are often interesting because they allow you to be more specific > about WHAT you want from someone. Actually, roles allow me to be more specific about what I want from someone. Realms just let me define where a user is defined. > Also, moves are afoot to better manage authentication information, which > IE already does, and Apple is introducing their KeyRings back into MacOS > to manage this. > > Part of the problem is sheer stupidity on e part of hte > implementations... for example, in Netscape, if you authorize to > "http://blah.com/Accounting/manage", and then try and access > "http://blah.com/manage", Netscape will automatically send your > Authorization header, when the standard specifically says: > > """ > A client SHOULD assume that all paths at or deeper than the depth of the > last symbolic element in the path field of the Request-URI also are > within the protection space specified by the Basic realm value of the > current challenge. A client MAY preemptively send the corresponding > Authorization header with requests for resources in that space without > receipt of another challenge from the server. > """ > > And in fact it shouldn't send it for higher hierarchical accesses. So, > to aleviate this problem, I think being able to separate the system into > multiple realms would be advantageous. Now, that doesn't mean that I > can't over-complicate things by saying that I think that acquisition > shouldn't be followed and that if user 'doe' is in the user folder of > the root object, which is realm "root" and tries to access a lower space > which is in realm "trusted" that she shouldn't be allowed access if > things are acquired that way. You should be able to look "up" the tree > and find things. So, if I understand you correctly, you want a user folder to implicitly contain users defined in user folders above, or, put another way, a user defined in a user folder is implicitly defined in any realms defined by user folders in sub-folders of the folder containing the user's user folder. I think this is a reasonable compromise. Having user folders, rather than ZPublisher do acquisition of user definitions is a direction that we've been moving in. > > > If the later, we find it convenient to have a "logout" > > document sitting around: > > > >  > > You have logged out<p> > >  > > > > if you really want subfolders to have their own realms, > > it probably wouldn't be too hard to > > provide a user folder property for the realm > > and have the user folder set the necessary response header > > when an authorization fails. > > Yes, this is kinda what I'm trying to get to. What effect does > Unauthorized have? It causes the browser to display a log-in box. Note that there is currently a problem being that in cases where Unauthorized exceptions are raised by application code, the realm is not set. We plan to fix this. Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

1 0

- I am *really* getting confused...
by skip＠calendar.com 08 Dec '98

08 Dec '98

(For those of you old enough to remember TV sets with rabbit ears, manual tuning and horizontal and vertical hold knobs, I feel like a little kid who has tried to fix the picture on the TV but has succeeded only in screwing up the picture further and finally had to give up and ask Dad to fix it...) Now I have this Product called calendar, which I can access as http://localhost:8080/Calendar/show Only problem is, it wants me to enter a username and password to access it. If I create a Document object named doc1, e.g. http://localhost:8080/doc1 I have no such problems. I created a sample Product named foo and was able to access it as http://localhost:8080/foo which worked, but it displayed with the [Properties|View|Security|Help] menu across the top of the displayed page. I've fiddled with the settings so much I think I may need to save my cal.py module and reinstall Zope from scratch just to get back to a reasonable configuration. I am really confused. Dad? Help? Skip Montanaro | Mojam: "Uniting the World of Music" http://www.mojam.com/ skip(a)calendar.com | Musi-Cal: http://concerts.calendar.com/ 518-372-5583

2 1

Re: [Zope] - Authenication problems when on webhost?
by Jimmie Houchin 08 Dec '98

08 Dec '98

On Tue, 08 Dec 1998 15:16:29 "Phillip J. Eby" wrote: >At 01:54 PM 12/8/98 -0600, Jimmie Houchin wrote: >>I haven't installed Zope yet. I have been reading the documentation and >>posts here. There have been a couple of post by Michael Grinder which have >>not been answered and concern me. Maybe they have not been answered because >>there isn't a good solution. If that is the case I prefer to know. >> >>For those of us who have to use a webhosting service we are at the mercy of >>what they (the webhost) will permit and what is required configuration >>wise by Zope. >> >>Are there any options for configuring for Authorization that can be done by >>those of us who use a webhosting service? >> > >There are several ISP's whose servers support HTTP_AUTHORIZATION, including >Hiway, RapidSite, and a number of regional telcos who outsource their >hosting through Hiway/RapidSite. Hiway and RapidSite run an Apache variant >known as "RapidSite/Apa-1.2" which was patched to pass through the >HTTP_AUTHORIZATION environment variable to CGI scripts. In their CGI >environments, all scripts run as the Unix user ID of the customer, so there >are no security issues with making HTTP_AUTHORIZATION available. <insert question> I just learned from my webhost today that: "any cgi script started is under your name, its automatic" I guess this would mean for my situation anyway, that all I need to do is ask whether or not HTTP_AUTHORIZATION is available? If not, would they? Is their a simple way to test if it is already (automatically) available without asking tech support? >This means that ZOPE should run virtually out of the box on a Hiway or >RapidSite >host. A couple of points to mention, though... Hiway does not allow you >to run your own servers, so ZopeHTTPServer is absolutely out. Since Hiway >servers are also periodically rebooted for various kinds of maintenance, >you cannot rely on such a process staying up indefinitely anyway. This >also means you need to be sure if you're using an LRP that it can >auto-start from the web. > >I know all this stuff because I work for Hiway/RapidSite and have >implemented ZPublisher-based internal applications there. I've switched >away from using my work e-mail address on the list, however, because I >don't want anything I say to be misconstrued as official statements on the >part of the company. Thanks for the reply. I hope any/all site(s) with excellent support for Python/Zope does well and prospers. :) Jimmie Houchin >

2 1

- Next problem - looking for dtml file in the wrong place...
by skip＠calendar.com 08 Dec '98

08 Dec '98

I'm inching along. Thanks to Michel Pelletier and Andy Kuchling, I now know my calendar module is a Product and what most (all?) of the necessary scaffolding is. My Calendar class supports a show method: def show(self, year, month=0): """return plain ASCII calendar month == 0 ==> display calendar for entire year """ if year < 1 or year > 9999: raise ValueError, ("year out of range: %d" % year) if month < 0 or month > 12: raise ValueError, ("month out of range: %d" % month) if month: cal = os.popen("cal %s %s" % (month, year)).read() else: cal = os.popen("cal %s" % year).read() return HTMLFile('cal', month=month, year=year, cal=cal) When I access http://localhost:8080/Calendar/show?month:int=0&year:int=1998 I get the following traceback: Traceback (innermost last): File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/ZPublisher/Publish.py, line 861, in publish_module File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/ZPublisher/Publish.py, line 585, in publish (Info: /Calendar/show) File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/ZPublisher/Response.py, line 284, in setBody File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/DocumentTemplate/DT_HTML.py, line 217, in __str__ (Object: /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/cal.dtml) File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/DocumentTemplate/DT_HTML.py, line 209, in quotedHTML (Object: /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/cal.dtml) File /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/DocumentTemplate/DT_String.py, line 532, in read_raw (Object: /home/dolphin/skip/src/Zope-1.9b1-src/lib/python/cal.dtml) IOError: (2, 'No such file or directory') which suggests to me that my cal.dtml file is not where Zope expects to find it. All files in my Products/Calendar directory are in that directory: % ls -F lib/python/Products/Calendar/ __init__.py cal.dtml vcal.gif __init__.pyc cal.py addCalendar_form.dtml cal.pyc Why is it looking up two levels in the directory hierarchy for the DTML file? As far as I can tell, I'm using the same calling conventions for HTMLFile objects as the MailHost and sample Products do. Any suggestions this time? Skip Montanaro | Mojam: "Uniting the World of Music" http://www.mojam.com/ skip(a)calendar.com | Musi-Cal: http://concerts.calendar.com/ 518-372-5583

3 4

- PROPOSAL: Rearrange Zope Documents
by Jim Fulton 08 Dec '98

08 Dec '98

"Christopher G. Petrilli" wrote: > > OK, this REALLY shouldn't be difficult, but I've tried every variation > of '_' and everythign else I can think of, read the source, seen > references to _p_mtime that don't seem to go anywhere... I'm obviously > missing something hideously transparent... > > How do I insert the time a document was last modified? > > i.e. I wanna create standard_html_footer that has something like: > > "Last modified: date here" > > at the bottom. I can't get anything to resolve correctly. The problem is that Documents are not what you think they are. They are miss-named. They should be called "DTML Methods". There should be another kind of object called "DTML Documents" that are meant to be the source of information, not a method. They should have properties and search their own attribute namespace before their folder's. Then, when a DTML method, like standard_html_footer, was used in a DTML document, and had:  the modification time of the DTML document would be inserted. I propose, by the end of the Zope beta period to: - Rename Document to "DTML Method", - Add a new object type, "DTML Document", and as a bonus: - Add a new object type, "Structured Text Document", that would use Structured text input and allow DTML tags to be interspursed, possibly using the Python string format or some other less-verbose format than SSI. Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

4 4

- Re: ExtensionClass/__of__ credit
by Jim Fulton 08 Dec '98

08 Dec '98

"Phillip J. Eby" wrote: > > Hey Jim, I'm in the midst of writing a paper on ASDF, and I want to > properly credit the invention of the '__of__' protocol and its > ExtensionClass implementation. I'm in the middle of a section where I'm > explaining how __of__ plus wrappers is the world's simplest (and coolest) > meta-object protocol for implementing acquisition, inheritance, and > method-to-instance binding. > > Currently I am referring to it as: > > "a simple, yet powerful dynamic binding technique invented by computer > scientist Jim Fulton (Digital Creations' Technical Director)" > > Is this correct? If you are looking for a reference I would use some think like: "The __of__ protocol was proposed by Jim Fulton and implemented in the ExtensionClass system [Fulton 96x] ... [Fulton 96x] Fulton, James L., Extension Classes, Python Extension Types Become Classes, http://www.digicool.com/releases/ExtensionClass/ExtensionClass.html, 1996. I happen to agree with you that __of__ an extremely useful protocol that is a natural extension of Python's method mechanism. I feel as though I've been a lone voice in the wilderness advocating it. Are you a member of the types-sig, http://www.python.org/sigs/types-sig/ ? That would be a good place to champion __of__. I think there's been some relevant discussion there, but I've been to busy with Zope to pay attention. :( Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

2 1

- iterative sorting w/dtml
by kent＠eaenki.nde.swri.edu 08 Dec '98

08 Dec '98

when using:  How can one filter the 'name' list prior to sorting? For example, to lower(case) the values, but I am looking for a more general solution.

3 3

- Importing objects
by Andy Dustman 08 Dec '98

08 Dec '98

I have an application where I would collect various automated e-mail responses which arrive more or less randomly. How can I get these into Zope? It looks like I could use the ZPublisher.Client module. Following along in the code, it looks like it ought to do the job. Is there some existing object type I can use for this, or do I have to make one? Note: the Client module uses rand, which I think is obsoleted; it is no longer in the Python Library Reference. To get around this, I did: #from rand import rand from whrandom import randint def rand(): return randint(0,2**31) Although, the only thing it is doing with rand.rand() is making a MIME multipart boundary, which could also be done with mimetools.choose_boundary(). -- Andy Dustman You should always say "spam" and "eggs" ComStar Communications Corp. instead of "foo" and "bar" (706) 549-7689 | PGP KeyID=0xC72F3F1D in Python examples. (Mark Lutz)

4 5

Re: [Zope] - Finding recently changed pages
by Jim Fulton 08 Dec '98

08 Dec '98

Kent Polk wrote: > > Hi A.M. (A.M. Kuchling), in <199812061624.LAA25965(a)207-172-39-232.s232.tnt10.ann.erols.com> on Dec 6 you wrote: > > > At http://discuss.userland.com/msgReader$857 , Dave Winer suggests > > that Web sites should have a top-level siteChanges.xml file that lists > > URLs that have recently changed. Search engines could then use this > > information to only crawl pages that have changed recently, which > > would let them keep up to date more easily. How would one go about > > implementing something like this under Zope? Andrew, I seem to have missplaced your original post, so I'll reply here. One problem we've run into whenever we've approached this is that it's really hard to *define* when a URL changes. If A URL is computed fron n objects, when does a URL change? When one of the n objects have changed? When they've all changed? When m (1 <= m <= n) have changed? When some particular subobjects have changed? I suspect that the definition could be application dependent. Jim -- Jim Fulton mailto:jim@digicool.com Technical Director (540) 371-6909 Python Powered! Digital Creations http://www.digicool.com http://www.python.org Under US Code Title 47, Sec.227(b)(1)(C), Sec.227(a)(2)(B) This email address may not be added to any commercial mail list with out my permission. Violation of my privacy with advertising or SPAM will result in a suit for a MINIMUM of $500 damages/incident, $1500 for repeats.

2 1

- Mailman Problems
by Amos Latteier 08 Dec '98

08 Dec '98

Hi Folks: We've been having a fair amount of trouble with Mailman. So if something isn't working right for you please bear with us; we're aware that there are problems and we're trying to fix them. Also: * We are currently running mailman behind a firewall so you CANNOT manage your subscription through the web. * I put instructions on managing your subscription with email on the Zope site: http://www.zope.org/Community/Mailman Thanks for your patience. -Amos -- Amos Latteier mailto:amos@digicool.com Digital Creations http://www.digicool.com

1 0