September 2000 - Zope

Re: [Zope] ZSQL
by Dieter Maurer 13 Sep '00

13 Sep '00

Tom Deprez writes: > ... nested Z SQL method call ... I do not understand fully, what you want to do. But the following FAQ may also be your problem: Z SQL methods do *NOT* use the DTML namespace for parameters. You must either pass the parameters explicitly: <dtml-in "sql_method(param1=val1, ..., paramn=valn)"> or put the parameters in the REQUEST object: <dtml-call "REQUEST.set('param1', val1)"> .... <dtml-call "REQUEST.set('paramn', valn)"> <dtml-in sql_method> Dieter

1 0

Re: [Zope] Defining a session idle timeout value using FSSession...
by Dieter Maurer 13 Sep '00

13 Sep '00

Aaron Bostick writes: > Would anyone on the list have a solution to implementing idle timeouts using > FSSession? > > I know you can cron up a simple job to delete the FSSession hierarchy based > on date and time, but how do you know which sessions are still being used if > you want to delete stuff say every 10 minutes or so? Under Unix, your code may look at the "last access time" (or universally the "modification time"). In Python, you can use the function "os.stat" to obtain this time. Dieter

1 0

_safe_getob error with LocalFS
by Michael Best 13 Sep '00

13 Sep '00

We are running: Solaris 2.6 Zope 2.1.6 LocalFS v0.1.4, later upgraded to v0.9.5 Revisions: 1.13 and 1.49 Using a simple dtml method on the local filesystem according to: http://www.zope.org/Members/jfarr/HowTo/DTML_with_LocalFS <dtml-var "files['foo.txt'].data"> Any ideas as to a solution to this problem? A traceback is listed at the end of this email. I am not directly on the Zope mailing list, but I do check the archives quite often, replies to the list and directly to my address would be appreciated. -Mike A quick list of all products installed: External Method Hotfix 06-16-2000 LocalFS MIMETools MailHost Navigator 1.02 OFSP PSQLInput PythonMethod 0-1-7 SiteAccess 1-0-1 Squishdot 0-3-2 TinyTable 0-8-2 ZCatalog 1-0-0 ZGadflyDA ZMySQLDA 1-1-3 ZNavigator 2.00 ZSQLMethods ------------------------------------------------------------------- error_type = AttributeError<br> error_value = _safe_getob<br> <pre> Traceback (innermost last): File /usr/local/Zope-2.1.6-src/lib/python/ZPublisher/Publish.py, line 165, in publish File /usr/local/Zope-2.1.6-src/lib/python/ZPublisher/mapply.py, line 160, in mapply (Object: dtmlwithexternal_test) File /usr/local/Zope-2.1.6-src/lib/python/ZPublisher/Publish.py, line 102, in call_object (Object: dtmlwithexternal_test) File /usr/local/Zope-2.1.6-src/lib/python/OFS/DTMLDocument.py, line 166, in __call__ (Object: dtmlwithexternal_test) File /usr/local/Zope-2.1.6-src/lib/python/DocumentTemplate/DT_String.py, line 502, in __call__ (Object: dtmlwithexternal_test) File /usr/local/Zope-2.1.6-src/lib/python/DocumentTemplate/DT_Util.py, line 335, in eval (Object: files['foo.txt'].data) (Info: files) File <string>, line 0, in ? File /usr/local/Zope-2.1.6-src/lib/python/DocumentTemplate/DT_Util.py, line 161, in careful_getitem File /home/teckno/zope/Products/LocalFS/LocalFS.py, line 447, in __getitem__ (Object: files) AttributeError: _safe_getob

1 0

Re: [Zope] Backslash as line continuation in C (Unix)
by Dieter Maurer 13 Sep '00

13 Sep '00

Hung Jung Lu writes: > ... I ran into troubles in > the compilation process because of the usage of backslash \ as line > continuation. (This is used in #def preprocessor macro statements.) When I > remove the backslashes and join the lines, it compiles fine.... > > Does anyone know any other better solution than go into the code and > explicitly remove the backslashes? Is there any gcc option that recognizes > backslashes are line-joining characters? Or is there another character used > in gcc for joining lines? Any other suggestions? Backslashes *ARE* line-joining characters for gcc -- without any options! I expect, that your backslashes are not immediately followed by "\n" (not "\r\n"!). Then, they would not be "line joining"! This very easily happens, if you transfer a Windows source onto a Unix system. I do not know, which editor you are using under Linux. If you use "vim", then you would not see the "\r" preceeding the "\n" but you would see a "textmode" in the status line. Use ":set notextmode" and then write the file back. This would remove the "\r". You may use the "tr" utility, too, for the "\r" removal, once you determined that this indeed is the problem. Dieter

1 0

shane@digicool.com
by Shane Hathaway 13 Sep '00

13 Sep '00

test. 1,2,3...

1 0

img tag dtml question
by Jason Cunliffe 13 Sep '00

13 Sep '00

Hello Can anyone see why under 2.2.0 this is wrong ? ............................................................................ /images [folder] 7Dlogoicon_gif [gif image file object] /deliverables/standard_html_footer [DTMLmethod] <dtml-with /images> <a href = /deliverables> <dtml-var "7Dlogoicon_gif.tag(border = 0)"> </a> </dtml-with> </body> </html> This worked many times before ok under 2.1.6 on another site with similar technique. But now I am getting >>> invalid syntax, for tag <dtml-var "7Dlogoicon_gif.tag(border=0)">, on line 3 of standard_html_footer When I try removing the outer quotes <dtml-var 7Dlogoicon_gif.tag(border = 0)> >>> Document Template Parse Error >>> invalid parameter: " = 0)", for tag <dtml-var 7Dlogoicon_gif.tag(border = 0)>, on line 3 of standard_html_footer Simply, I need to make sure that every document I put in '/deliverables' can have a localized standard HTML footer with an icon which will link users back up to /deliverables. The icon needs to have a 'border = 0' attribute. help much appreciated thanks - Jason ________________________________________________________________ Jason CUNLIFFE = NOMADICS.(Interactive Art and Technology).Design Director

1 0

Ownership and security
by Andy McKay 13 Sep '00

13 Sep '00

How do I remove ownership from a product, ie: make it unowned? I have some dtml that works quite happily on unowned objects, but at soon as they become owned I get into a whole whack of problems with getUserById. Of course I would really love to have ownership work, but I know its hell trying to explain it on a mailing list. -- Andy McKay, Developer. ActiveState.

1 1

Re: [Zope] How to Pass values to a DTML method ??
by Dieter Maurer 13 Sep '00

13 Sep '00

Evan Simpson writes: > .... extended-attribute syntax > (foo-name-name="x" instead of foo="_[_[x]]"). HELP! I am sure, that such an extended attribute syntax would cause hundreds of messages in this list. I hope, it will not get adopted. > I have considered proposing that we graft <dtml-let>'s capabilities onto the > other tags, so that we could write stuff like: > > <dtml-var set-foo="getFoo()" var=foo set-param1=" 'text' " set-param2="id"> That is better. Only: What does the "var=foo" mean? The thing that should be rendered? Then (with respect to this), the current syntax is clearer. Or: Should this be equivalent to: <dtml-var "getFoo(_.None,_,param1='text', param2=id)>? Then, it is not at all obvious. "getFoo" seams to be called explicitly without parameters. Clearer would be: <dtml-var getFoo set-param1="'text'" set-param2="id"> Dieter

1 0

GPL stuff (OT)
by Phill Hugo 13 Sep '00

13 Sep '00

Zopistas, I've heard a lot of GPL issued raised recently and I want to say a couple of things to hopefully end the subject. I don't mind you pointing replies to me if they aren't of general interest [hint ;)] Firstly, read the GPL. It sounds like some of you talking about it haven't (or have forgotten its contents). That'll clear most things up. Its ambiguousity is limited to very technical cases such as dynamic linking etc. Even in that case, the use of the GPL'd C Header files at compile time can be taken to be indication of agreement to the GPL (even the use of header files produces a legally defined "derivative work" - one look at the sql.h files MS distibute to make that clear. The lack of court testing is irrelevent too, that hasn't happened becuase no one has stood to gain from it - a good indication that they'd lose. If any company found real legal fault with the GPL where they could take code and not be force to comply, they would have done. The legal issues surround very specfic cases, not the entire license. The Objective C front end to GCC is there because NeXT backed down after admitting a futile case. The "its not been tested in court" argument applies for many many licenses and EULAs but you can bet that the flip side is true also - the reason its not got to court is becuase everyone seems to be agreeing to it out of court - a precedent no court would ignore if it ever came to it. The compatibility issues of [MQZ]PL / GPL are generally cases where some case is strongly demanded which the GPL strongly rejects. Things like choice of law, the right to take patches for non Free use etc would cause incompatibility due to the difference in the rights assigned to the user under GPL and those not allowed under the other. This DOESN'T stop you using the GPL in Zope though if you specifically allow it on your side and the creators also do. As far as I know, DC hasn't forbade modules under GPL to be used, or rather they have made the Product interface "open" enough to allow even proprietry Products to be written and used. Of course, this is module authoring (even Linux kernel modules need not be under GPL). In the case of taking ObjectManager.py and adding things to it, that stays under ZPL like it or not (even your changes). That is where the two are incompatible. You can take embed GPL code in BSD code - that's why they are "compatible". To call that "pilfering" but not mention companies who take BSD code and don't even share their extenstions is a little harsh I think (the windows FTP command comes to mind). I don't think that using a GPLd product in Zope (via the Product interface) would mean you have to give source away for _your_ Products and site code which uses them. They are distinct and bound only by the open interface Zope provides. If, however, you change the GPL'd Product, then you will have to stick with the GPL. Either that or re-implement. And yes, you will have to offer the GPL'd parts under the GPL terms if you distribute the whole. A real world case of this? There are many commercial products that include GPL code. I've seen it used in firewalls and printers and in all cases the GPL parts where distibuted under GPL of course. The distinct parts (non linked, seperated by "open interface" - be that as seperate processes, kernel modules etc) are not always available under GPL. Philips' Tivo site has their changes to Linux under GPL available. You won't find their MPEG suite though - shame ;) The GPL has no direct stance on making money or otherwise. You can sell GPL'd code for all you like, changes or none. It only governs the rights assigned to those YOU distribute it to (not the world). However, that one customer then has the right to distribute to the world. Unless you are bloody good with your support and documentation, they won't need you. If you are good that's a different matter. The choice of license on an open project isn't, from the real world, an indication of how well it'll do commercially either. Yes, apache, XFree86, Perl, Python etc etc all use non GPL licenses and that's good for them in some cases - their aims are DIFFERENT to FSF/GNUs. Besides, Linux, GNOME and the GNU compiler tools get just the same amount of commercial recognition (Mozilla is now also under GPL), if not more and from just the same people. That's about it. The only really clarifcation needed to clear this all up is... 1) DC to say Products do not have to be under ZPL and to deem their Product API an open interface akin to Linus' statement on the Linux kernel module interface or RMS's on calling GPL'd apps through the System() call. 2) Any GPL code that effects Zope's core will have to be deemed under ZPL while running with Zope. I'm not sure any Products do this but in the case of any, this would sort it out. Yes, its really just a dual licensing scheme but many do that - Mozilla, Perl and Qt amongst them (and all with the GPL alongside their other license). In the case of the first being granted, its only the latter that would cause "distribution concerns" I think. Thankfully its also a lot more unlikely to happen. Phill phill(a)gnu.org

1 0

Zope 2.2.1 Vrs Apache Benchmarks using Apache Bench (AB)
by J. Atwood 13 Sep '00

13 Sep '00

I have done this before (http://www.zope.org/Members/BwanaZulia/benchmarks.html) but I just got my hands on a brand new server and thought I would see how 2.2.1 does. Machine: 600 MHz, 128 MB Ram, 20 GB EIDE drive, Linux 6.2 Ran it 5 times.. always came out close to this (+/- 1 RPS) Server Software: Zope/Zope Server Port: 80 Document Path: /index_html Document Length: 184 bytes Concurrency Level: 25 Time taken for tests: 9.715 seconds Complete requests: 1000 Failed requests: 0 Total transferred: 387387 bytes HTML transferred: 184184 bytes Requests per second: 102.93 Transfer rate: 39.88 kb/s received Connnection Times (ms) min avg max Connect: 0 5 254 Processing: 100 234 36 Total: 100 239 290 And Apache which I ran a lot. The RPS were anywhere from 250-850.. so this was about average of all of them. Server Software: Apache/1.3.12 Server Port: 8080 Document Path: /index.html Document Length: 231 bytes Concurrency Level: 25 Time taken for tests: 1.896 seconds Complete requests: 1000 Failed requests: 0 Total transferred: 520520 bytes HTML transferred: 231231 bytes Requests per second: 527.43 Transfer rate: 274.54 kb/s received Connnection Times (ms) min avg max Connect: 0 3 27 Processing: 18 42 706 Total: 18 45 733 Interesting as I upped the anti (requests and concurrency) Zope stuck to about 95-100 RPS and Apache started to lower (250-650). Disclaimer: This was a totally static test and it only shows how fast Zope can serve up static pages as Apache certainly does not have all the features of Zope. This is not very scientific just interesting. Take a look the page linked at the top if you want to have look at some other tests. J

1 0