- Zope-Dev - Zope lists

[ANN] Zope 2.2 patch for cDocumentTemplate/Acquisition bug
by Dieter Maurer 05 Nov '00

05 Nov '00

Appended is a patch for the cDocumentTemplate/Acquisition bug. Usually, this bug manifests itself by an "AttributeError: __call__" for a folder object. The usual context is: DTML Document: ..... <dtml-with folder>  .... </dtml-with> ..... The patch modifies "C" code. This means you need a Zope source distribution to apply it and then you must regenerate Zope. Almost surely, the patch is not yet complete: The code in "cDocumentTemplate.c" looks like (Python emulating "C" code): if callable(o): if hasattr('isDocTemp') and o.isDocTemp: o= o(None,md) else: o= o() The remaining problem is that 'isDocTemp' might be acquired and does not belong to "o.__call__". In this case, "o" might be called with the wrong number of parameters. I would expect, that the problem disappears, when the "isDocTemp" test is performed with "o.aq_base" rather than "o". But I am not sure. Maybe, an acquisition expert can look into this remaining issue. Dieter ------------------------------------------------------------------------ --- lib/Components/ExtensionClass/:Acquisition.c Sat Oct 7 20:40:53 2000 +++ lib/Components/ExtensionClass/Acquisition.c Sat Nov 4 21:10:04 2000 @@ -1391,6 +1391,31 @@ return r; } + +/* DM: adding "AqCallable_Check" */ +static int +aqCallable(PyObject *o) { + if (isWrapper(o)) { + PyObject *call; + + UNLESS(call=PyObject_GetAttr(o,py__call__)) { + PyErr_Clear(); + return 0; + } + Py_DECREF(call); return 1; + } else + return PyCallable_Check(o); +} + +static PyObject * +AqCallable_Check(PyObject *ignored, PyObject *args) +{ + PyObject *o; + + UNLESS (PyArg_ParseTuple(args, "O", &o)) return NULL; + return PyInt_FromLong((long) aqCallable(o)); +} + static struct PyMethodDef methods[] = { {"aq_acquire", (PyCFunction)module_aq_acquire, METH_VARARGS|METH_KEYWORDS, "aq_acquire(ob, name [, filter, extra, explicit]) -- " @@ -1412,6 +1437,10 @@ {"aq_chain", (PyCFunction)module_aq_chain, METH_VARARGS, "aq_chain(ob [, containment]) -- " "Get a list of objects in the acquisition environment"}, + /* DM */ + {"callable", (PyCFunction)AqCallable_Check, METH_VARARGS, + "callable(ob) -- " + "return true iff the object is callable"}, {NULL, NULL} }; --- lib/python/DocumentTemplate/:cDocumentTemplate.c Sat Jul 29 11:24:34 2000 +++ lib/python/DocumentTemplate/cDocumentTemplate.c Sat Nov 4 21:11:11 2000 @@ -95,14 +95,35 @@ static PyObject *py_Unauthorized_fmt, *py_validate; static PyObject *py__push, *py__pop, *py_aq_base; -/* ----------------------------------------------------- */ - static void PyVar_Assign(PyObject **v, PyObject *e) { Py_XDECREF(*v); *v=e;} #define ASSIGN(V,E) PyVar_Assign(&(V),(E)) #define UNLESS(E) if (!(E)) #define UNLESS_ASSIGN(V,E) ASSIGN(V,E); UNLESS(V) #define OBJECT(O)(((PyObject*)O)) +/* ----------------------------------------------------- */ +/* DM: callable */ +static int (*callable)(PyObject *o); + +static PyObject *aq_callable; + +static int aq_callable_check(PyObject *o) { + PyObject *result= PyObject_CallFunction(aq_callable,"O",o); + int callable= PyInt_AsLong(result); + Py_DECREF(result); + return callable; +} + +static void init_callable() { + PyObject *aq= PyImport_ImportModule("Acquisition"); + if(aq && + (aq_callable= PyObject_GetAttrString(aq,"callable"))) + callable= aq_callable_check; + else callable= PyCallable_Check; +} + +/* ----------------------------------------------------- */ + typedef struct { PyObject_HEAD PyObject *inst; @@ -370,7 +391,7 @@ { dt=0; - if (PyCallable_Check(e)) + if ((*callable)(e)) { /* Decide whether we have a document template */ if (rr=PyObject_GetAttr(e,py_isDocTemp)) @@ -917,6 +938,9 @@ PyDict_SetItemString(d, "__version__", PyString_FromStringAndSize(rev+11,strlen(rev+11)-2)); + + /* DM */ + init_callable(); if (PyErr_Occurred()) Py_FatalError("can't initialize module cDocumentTemplate");

1 0

Re: [Zope] Question about dtml-in
by Dieter Maurer 05 Nov '00

05 Nov '00

Sean McGrath writes: > ... "dtml-in" treats sequences of two element tuples specially ... You are right. This is to support the frequent case of iterating over item sequences. There should be a "dtml-in" argument to switch off this special treatment. If I had your problem, I would go for such a solution (extending "dtml-in"). It should be easier than to work around the special treatment in DTML and it would help the Zope community as a whole. If you do not want to go this way, I would suggest to put a feature request into the Collector. Dieter

1 0

Re: [Zope-dev] ZCLass help
by Dieter Maurer 05 Nov '00

05 Nov '00

Veiko Schnabel writes: > .... > i created a new Product calles book > with propertys writer(string), pages(int) > and a standart method called index_html to output all these > probertys .... > .... instance "my_book" .... > <dtml-var my_book> > .... > all i get is the id of the instance my_book > <PubClass instance at 8ab30c8> The ZPublisher (i.e. when you access your instance with the internet) automatically activates "index_html" (if the object is not callable). "dtml-var" does not. Instead, it calls the object, if it is callable (not the case in your example), and then converts to a string. You can use: <dtml-with my_book> <dtml-var index_html> </dtml-with> Dieter

1 0

Re: [Zope-dev] another debugging howto, and a question
by Dieter Maurer 05 Nov '00

05 Nov '00

Simon Michael writes: > Here are some draft howto-like things some might find helpful > (improvements welcome) : > > http://zwiki.org/ZopeDebuggingRoadmap > http://zwiki.org/HowToDebugZopeInEmacs > > and a burning question: > > how can I invoke zserver such that I will get the interactive debug > prompt ? I'd like to have my breakpoints trigger when I access the > server in the normal way with a web browser, poke around, let it run > on and do it again. I was able to do this in the past and it's useful. Maybe, the monitor is what you need. In a monitor session, you can import pdb. However, as Zope is multi-threaded, pdb interaction might seriously affect the other threads. To put it another way: I am not sure, whether pdb is thread safe. Dieter

1 0

python memory error on Windows NT
by James Johnson 04 Nov '00

04 Nov '00

Error messages are wonderful things aren't they? If you notice other quirky/unexplained behavior with the machine, it could be a RAM problem. You say you have 2 ODBC connections and 2 Access databases. Are they located on the same machine or across a LAN? One hunch is that the problem is that because of network traffic, complex queries, proxy access. Some kind of delay however slight could be causing the ODBC software to fail. The RAM does seem suspect and some other process on your machine could be set to aggressively take RAM space. I would look into the way Zodbc handles network latency and error conditions between 2 or more ODBC databases. I wonder from what I just suggested would it be possible to get this error from Zodbc? python.exe >The instruction at "0x1b271a1f" >referenced memory at "0x00000000" >The memory could not be "read" Get your Free E-mail at http://tacoma.zzn.com ____________________________________________________________ Get your own Web-Based E-mail Service at http://www.zzn.com

1 0

Re: Zope-Dev digest, Vol 1 #785 - 21 msgs
by Colin Leath 04 Nov '00

04 Nov '00

I get that error (replace "python.exe" with the program that just crashed) somewhat often on my NT Server box at work. I notice it most often when using Excel, or when working in Installshield. I assume one of my memory chips has a problem... If you find a different solution, please let me know. Colin cleath(a)experienceartist.org Message: 20 From: "Coi Giovanni" <coi.giovanni(a)prometeo.it> Organization: Prometeo srl To: Toby Dickenson <mbel44(a)dial.pipex.net>, tdickenson(a)geminidataloggers.com Date: Fri, 3 Nov 2000 17:10:23 +0100 Subject: Re: [Zope-dev] python memory error on Windows NT Reply-to: coi(a)prometeo.it CC: zope-dev(a)zope.org In data 3 Nov 2000, alle ore 11:22, Toby Dickenson ha scritto: > >someone has never seen this kind of error on Windows NT > >4.0 (last patch update) while executing IIS 4.0 + Zope 2.2.2? > >The panel say: > > > >python.exe > >The instruction at "0x1b271a1f" > >referenced memory at "0x00000000" > >The memory could not be "read" >

1 0

Re: [Zope] pluggable brains
by Philipp Auersperg 04 Nov '00

04 Nov '00

With that I can implement some business logic for the merchant objects in an object oriented manner, so these objects know how to handle themselves. read the article in (http://www.zope.org/Members/michel/ZB/RelationalDatabases.html) about object-relational binding, there you find some more detailed descriptions but I think it is still a little bit underdocumented... BTW in meanwhile I've found the answer to my question below: you can just bind ZClasses to a SQL method that are not marked as persistent. phil *********** REPLY SEPARATOR *********** On 03.11.2000 at 14:11 MANUEL JOSE AMADOR BRIZ wrote: >What is this ZClass binding for? What\'s its purpose? > >Quoting Philipp Auersperg <zope(a)philosoft.at>: > >> For ZSQL Methods it is possible to bind a ZClass to the records. >> >> These can be selected from te combo box labeled \'ZClass\' in the >> \'advanced\' tab of the SQLMethod\'s management screen, >> but there is always listed just one ZClass. >> >> My question: >> Why is there just one ZClass selectable, even if there are many ZClasses >> defined? >> Is that a bug or a feature? >> >> thanks >> >> phil >> >> > >---------------------------------------------------------- > Universidad Federico Santa Maria - Campus Guayaquil > Enviado por: www.usm.edu.ec

1 0

2gb file size
by Andy McKay 04 Nov '00

04 Nov '00

My database has suddenly ballooned to 2 gig's stopping my Zope from running. Are there any utils to delete the last transaction or so? This is kinda urgent... :( Thanks. -- Andy McKay, Developer. ActiveState.

2 1

New proposal - WikiForNow
by Ken Manheimer 03 Nov '00

03 Nov '00

A week or so ago i threatened to post a proposal for near-term fixes of wikis on zope.org to alleviate some of the glaring fishbowl pain (looking to collect preliminary suggestions) - well, it's here: http://dev.zope.org/Wikis/DevSite/Proposals/WikiForNow >From the problem statement: While wikis as they stand provide benefits for the fishbowl process, we've all been chafing under some shortcomings, some of which need to be rectified soon, to reduce chafing and increase scalability of the process to increasing numbers of items (proposals and projects)... Feedback is welcome, here in zope.dev and on the discussion page: http://dev.zope.org/Wikis/DevSite/Proposals/WikiForNow If you give feedback that has a conclusion here in zope.dev, realize that it may slip by the way until you register it in the discussion page. Conversely, if you respond there, please post a note here so concerned parties (including me) know to check them out. (This notification awkwardness is one of the things i think important to address soon, hence part of the proposal...) -- Ken Manheimer klm(a)digicool.com

1 0

cPickle.c
by Robin Becker 03 Nov '00

03 Nov '00

I see a reference to mymath.h in in cPickle.c is that intentional? -- Robin Becker

2 2