- Zope-Dev - Zope lists

first release of repozo.py available (was Re: [Zope-dev] simple revision control(?) for Data.fs.)
by Anthony Baxter 12 Jan '00

12 Jan '00

ok, a first release of repozo.py is up at http://www.zope.org/Members/anthony/software/repozo.py This makes backups of your ZODB (Data.fs) file when run with -S, only storing the full file when necessary. When run with -R it will recover the data.fs file as at the specified date. Run with -v to have it talk to you and tell you what it's doing and why. More docs to come. caveat emptor: this sucker is really very very new - and the code's pretty grotty. but I have to go home now, so I thought I'd fling out a quick release. Usage: -S --store backup current ZODB file -R --recover restore stored ZODB file Flags for --store and --recover: -z file --zodb=file ZODB file (default var/Data.fs) -r dir --repository=dir repository directory (default /opt/zope/saveDir) -v --verbose verbose mode - say what's happening Flags for --store: -F --force force a full ZODB backup Flags for --recover: -D str --date=str recover state as at this date. str in format yyyy-mm-dd[-hh[-mm]] -o file --output=file write recovered ZODB to file (default Data.fs.recover) Current todo list: . support gzipped files . filter non-repozo files from the repository . store files in a hierarchy of directories (by date) rather than one huge directory

1 0

Re: [Zope-dev] simple revision control(?) for Data.fs.
by Anthony Baxter 12 Jan '00

12 Jan '00

>>> "Stuart 'Zen' Bishop" wrote > A 'quick hack' method of doing this would be to ascii encode your > data.fs (uuencode, base64 - anything as long as you have line feeds) > and use RCS to handle all the deltas. I thought about that. It's ugly, and will consume even more space than the current ZODB. In the case of a pack, won't RCS just store the whole thing in the RCS file again? or will it use some other format? You're also going to end up with a huge Data.fs.uu,v (or whatever) file. > Or you could buy a Netapp and use the built-in snapshot facilities :-) :-P Anthony

1 0

ASP vs Zope vs PHP vs ..?
by Michael Waddell 12 Jan '00

12 Jan '00

I'm an ASP/VB/Site Server developer looking for an alternative. Zope is very appealing -- it looks like a great tool supported by a great community. And I've enjoyed looking at and playing with Python. But, I'm uncomfortable with loosing touch with an actual HTML/ASP document in an actual file system. I feel weird about the URL that Zope presents - there's no simple file name reference, like, http://www.abc.com/mypage.asp or http://www.abc.com/mypage2.html. Sorry, but my web experience expects that. 1. Where can I learn how others have handled the transition from page-based publishing to Zope's object-based paradigm? 2. How do you create/maintain the look of the site? Doesn't seem feasible using the Zope control panel. I prefer using HomeSite (Allaire) for coding HTML, ASP and any text file. MacroMedia DreamWeaver for complex HTML design/layout. And I like using MS SQL Server (although I could be converted..) to support data-driven web sites. How do I integrate these products with Zope (if at all). The HTML page editing/layout features of Zope are not clear to me. Thanks for aiding in my evaluation. Regards, Michael Waddell Michael Waddell MW Data Basics, Inc. 153 Boles Road Marshfield MA 02050

2 2

simple revision control(?) for Data.fs.
by Anthony Baxter 12 Jan '00

12 Jan '00

Hiya, I plan to write something that will do 'daily' (or whatever) archiving of Data.fs, by using a fairly simple algorithm: start: store file, size incremental: if new_size < size: store complete new file, size # pack operation has happened. else: store all bytes from size to new_size This should allow me to reconstruct the Data.fs at any given point... can anyone see any problems with this? (of course, the bonus followup would be to implement the CVS pserver protocol so that you could retrieve it that way, but hey, I've got better things to do with my life :) Anthony

3 4

Authentication Patch
by Evan Simpson 12 Jan '00

12 Jan '00

I got tired of dealing with the weird mechanics cookie-based logins have to deal with, where standard_html_error and tracebacks intrude on login intercepts. I came up with the following minor change, which works great for me, and might make GenericUserFolder's job easier, too. In ZPublisher/HTTPRequest.py (or SiteAccess/ChangeBehaviors.py, if you use SiteAccess), change the lines: if user is None and roles != UNSPECIFIED_ROLES: response.unauthorized() to: if user is None and roles != UNSPECIFIED_ROLES: object = response.unauthorized Under normal circumstances, this has no effect, since 'response.unauthorized' will be called as soon as it is returned and will raise its exception as usual. Now however, we add a twist; When the cookie-based authenticator fails, it replaces 'response.unauthorized' with its 'login' document. If no higher-level authentication succeeds, the 'login' document is rendered normally. Since 'response' is re-created with each request, this modification has no effect beyond the current request. What do you folks think? Cheers, Evan @ 4-am

1 0

Factory pollution, scopes, zope hosting
by Lalo Martins 12 Jan '00

12 Jan '00

I already described a problem of ``factory pollution'' I'm facing. In short, I develop a ZClass for a very specific problem, and then have to cope with its metatype in the ``add object'' lists of the entire site. This is a pain for hosting providers for two reasons: 1: it pollutes the ``add object'' list. I will soon have a half hundred metatypes. It's bad to make customers understand this. 2: perhaps I developed some class for one customer alone. For the others it's a waste of mental space. Perhaps I even want to charge for specific services (like, say, a shopping cart class). I was going to settle for deleting the Factories and making some scripts to add all instances programatically, but then I realized there is a possible sollution. I'd like to hear comments on it from folks with a say on Zope internals before I move on. This sollution is giving a ``scope'' property to the Factory object. This property is a list of strings (represented as ``lines'') defaulting to ['/']. A given factory is only valid in places where the path start with one of the scope strings (so the default scope says this factory is valid anywhere, but I could change it to ['/BroDar/nucleos', '/lalo/foobar'] so it would stop all other places from having this metatype on their ``add object'' lists). Is it worth? Is it a good sollution? Is the problem real, or in my mind? []s, |alo +---- -- I am Lalo of deB-org. You will be freed. Resistance is futile. http://www.webcom.com/lalo mailto:lalo@webcom.com pgp key in the web page Debian GNU/Linux --- http://www.debian.org Brazil of Darkness -- http://zope.gf.com.br/BroDar

1 0

Re: [Zope-dev] simple revision control(?) for Data.fs.
by Anthony Baxter 12 Jan '00

12 Jan '00

>>> Evan Gibson wrote > What if someone packed their database and then did heaps of changes so that > the size was bigger than it was before the packing? > You'd have to add extra checking that, if it was bigger, the first "size" > bytes are the same in both files. ooo. nasty. hm, ok, so I need to check for the datestamp on Data.fs.old Or maybe md5sum the bytes... hmmmm. Anthony

1 0

RE: [Zope-dev] RE: More on the Absent id bug
by Michel Pelletier 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Jason Spisak [mailto:jason@mtear.com] > > I commented out that line, and it still has the backslash in > front of the > catalogued url. > > Michel, is that right? > > Aren't they supposed to not have the leading slash? If it > has a leading > slash, then when you click on the link in the Catalog > management interface, > you get and error. (it is looking for a server @ > http://my_object/url/object, rather that > http://localhost:8080/my_object/url/object. > > Which one is correct in in your design? The latter, obviously. Not putting the slash in confuses your browser into thinking it's a bad relative URL. The culprit is in lib/python/Products/catalogView.dtml Here is the checkin message: ---------------------------- revision 1.10 date: 1999/12/28 23:59:10; author: amos; state: Exp; lines: +1 -1 Fixed links to catalogged objects. ---------------------------- Here is the diff: *** /tmp/catalogView.dtml.~1.9~1NAywY Tue Jan 11 11:24:02 2000 --- /tmp/catalogView.dtml1NA_6e Tue Jan 11 11:24:02 2000 *************** *** 48,54 **** </TD> <td valign="top"><dtml-var meta_type></td> <td valign="top" align="left"> ! <a href="/<dtml-var "getpath(data_record_id_)">/manage_workspace"><dtml-var "getpath(data_record_id_)"> <dtml-if title> (<dtml-var title>)</dtml-if></a> </td> </tr> --- 48,54 ---- </TD> <td valign="top"><dtml-var meta_type></td> <td valign="top" align="left"> ! <a href="<dtml-var "getpath(data_record_id_)">/manage_workspace"><dtml-var "getpath(data_record_id_)"> <dtml-if title> (<dtml-var title>)</dtml-if></a> </td> </tr> As you can see what probably happened was that not slash before the <dtml-var getpath...> probably broke the catalog's ability to show good links in certain circumstances, which fixed one problem but manifested yours. The quick fix is to remove the / in the <a> tag in this file. But we MUST think more about this path stuff, Catalog was the first app to use it extensivly and our lack of a path framework shows. -Michel

3 2

RE: Uncatalog absent id bug - one possible cause
by Michel Pelletier 11 Jan '00

11 Jan '00

> -----Original Message----- > From: Itamar Shtull-Trauring [mailto:itamars@ibm.net] > Sent: Tuesday, January 11, 2000 7:18 AM > To: zope-dev(a)zope.org > Cc: michel(a)digicool.com > Subject: Uncatalog absent id bug - one possible cause > > > Good morning Michael, > > I'm also encountering the "Uncatalog absent id" in a bad way > (users can't > enter data cause reindex fails), and I found at least one > reason why, in > addition to the slash problem (which I deal with by trying to > uncatalog both > with and without a slash.) > > I have a folder with a space in it - "Do Ahead". When I > index using the > "Find objects to index", they get indexed as "/Do > Ahead/object". When I > try to unindex them it tries to unindex "/Do%20Ahead/object", > and then that > error comes up. I guess here is a good example of where it makes sense to raise an error, otherwise it would have silently *not* unindexed your object. > > I added it to the Collector. Thanks, I'm guessing it's a simple fix of adding url_quote somewhere. -Michel > > -- > Itamar S.T. itamars(a)ibm.net >

1 0

RE: [Zope-dev] RE: More on the Absent id bug
by Michel Pelletier 11 Jan '00

11 Jan '00

I'm against doing this off hand without thinking about the problem, according to Jason this never happened before a certain recent time, but neither this code nor any code in Catalog changed at that time. This sort of makes me think this is not the problem. How is it, exactly, that this problem manifests itself for both of you? How are your objects getting cataloged? What changed when the problem started apearing? -Michel > -----Original Message----- > From: David Kankiewicz [mailto:kankie@thegrid.net] > Sent: Monday, January 10, 2000 7:09 PM > To: Michel Pelletier > Cc: 'Jason Spisak'; zope-dev(a)zope.org > Subject: Re: [Zope-dev] RE: More on the Absent id bug > > > Michel Pelletier wrote: > > > <cut> > > > > > Can you open a collector ticket? I'm exploring catalog issues right > > now... > > > > -Michel > > > > OK OK... About the leading slash ////// > > In ZCatalogAware.py > in function url() > > We have: > ... > if uri[0]=='/': uri=uri[1:] > return uri > > > If I comment out the "if", the ZCatalogAware subclassed > Zclasses catalog > themselves correctly, like Find's "/..." :) > > Now, what I want to know is, will this somehow mess up something else? > (I can't see how it would) If not PLEASE check it in to the cvs! > > Bye for now, > David >

1 0