*** sss has left #gnuenterprise
*** reinhard has joined #gnuenterprise
*** johannesV has joined #gnuenterprise
*** btami has joined #gnuenterprise
*** reinhard has quit IRC
*** reinhard has joined #gnuenterprise
*** sjc has joined #gnuenterprise
*** sjc has quit IRC
*** dimas has joined #gnuenterprise
<johannesV> damn, i've indroduce a new bug into appserver ... (will fix it right now)
<johannesV> phu, ...
*** kilo has joined #gnuenterprise
<kilo> good morning
<johannesV> hi kilo
<kilo> hi johannesV
<johannesV> kilo, what about your recursion problem (in inv_head.gfd)
<johannesV> is it still there ?
<kilo> yes, definitely
<kilo> lately the form has been simplified quite a bit, no free block, etc.
<kilo> and with that simplification the recursion problem is gone
<kilo> but another one rises...
<johannesV> which one ?
<kilo> the problematic area is still the trigger that wants to calculate the dateDue
<kilo> if it is enabled, the first record of the detail part (ie the invoice lines) disappears, only the second one gets stored...
<johannesV> can i reproduce this with current svn ?
*** dcmwai has quit IRC
<kilo> yes
<johannesV> how would i do this ?
<kilo> trigger on line 86, uncomment it
<kilo> then start form, click on dateIssued (to get the recent date, this is a bug also)
<kilo> then chose payment method, it computes dateDue
<kilo> then try to add first line, after you step out of that onto second one, the irst one disappears
<kilo> s/irst/first
<johannesV>     return self._cachedRecords[record]
<johannesV>  IndexError: list index out of range
<johannesV> (last two lines of exception)
<johannesV> is this the error you're talking of ?
<kilo> you get an exception???
<kilo> i don't get any, the line simply disappears
<johannesV> hmm
<johannesV> btw. you could simplify your procedures a bit ... :)
<johannesV> instead of:
<johannesV>       total = 0.0
<johannesV>       if len(items):
<johannesV>         #total = sum(items)
<johannesV>         for a in items:
<johannesV>           total += a.INV_valueGross
<johannesV> you could do:
<johannesV> total = sum ([float (i.INV_valueGross) for i in items])
<johannesV> (but i haven't tested this yet)
<johannesV> same applies to all that INV_value*
<kilo> hmm, i have thought about using some python iter funcs...
<johannesV> calcDateDue: why does it return None but set self.dateDue ?
<johannesV> why doesn't it return the same value as self.dateDue ?
<kilo> dunno...
<kilo> btw why use 'float' in the above code???
<johannesV> you could leave that away (it's used because of your printf-debug below)
<kilo> btw this form is a dialog. this way it does not show the dateIssued at first, only if you click on that entry. if you remove the 'style="dialog"' from the form head, it does show it...
<kilo> there must be something done differently on form startup depending on its dialog state
<johannesV> hmm, so think i can imagine where the bug is
<johannesV> as i've mentioned a few times ago: the difference between form and dialog is: in a form the program continues with normal execution after showing the window and a dialog holds program execution until the window get's closed
<johannesV> so after showing the dialog's window no more GF* events are processed which are issued *after* the window has been activated
<johannesV> we have to dig into form's code to find out how to circumvent this wrt initial values
<kilo> yes, i agree
*** lekma has joined #gnuenterprise
<lekma> morning (?) everyone
<kilo> morning lekma
<btami> mornin'
<btami> :)
*** ajmitch__ has quit IRC
<lekma> hi btami kilo
<reinhard> johannesV, kilo: now is there a bug you are working on currently?
<reinhard> or is everything ready for new release?
<kilo> i don't know if johannesV is working on it or not
<kilo> but there is a bug in Forms
<reinhard> ok
<reinhard> the i'll wait
<kilo> don't forget to release designer...
<reinhard> yes
<reinhard> i think it will be a complete round this time
<kilo> reinhard: included bug in forms/bugs
<reinhard> kilo: thanks
<reinhard> kilo: I'll let johannesV decide if he wants to look into it or not
<reinhard> if not, then I'll do prereleases today
<reinhard> if he wants to, I'll wait for the fix
<kilo> ok, don't bug his siesta...
*** holycow has joined #gnuenterprise
<btami> l8r
*** btami has quit IRC
* johannesV back
<johannesV> so my siesta is finished :)
<johannesV> kilo, i'm not sure if i can fix the dialog-thingy that quickly
<johannesV> as to the exception raised in INV_Head.gfd it seems like the bug i've filed in common last week
<johannesV> (but i haven't looked at it very close)
<kilo> np, it is included in frms/bugs
<kilo> strange about exception that you get it, while i do not...
<reinhard> ok so I do prereleases now
<johannesV> well, maybe my backend isn't very clean
<johannesV> reinhard, wait i'll give ya a call first
<johannesV> kilo: the date issued is now there from startup of the form ..
<johannesV> ah, you've changed the form-style ..
<johannesV> ok
<kilo> yes, you were quicker...
<kilo> do you like that goat?
<kilo> goat also disappears if the form-style is dialog...
<johannesV> right
<johannesV> ok
<johannesV> i've solved the dialog-prob
<kilo> that was quick
<reinhard> johannesV: is it in common or in forms?
<johannesV> forms
<johannesV> well, actually i've moved on single line
<johannesV> but i need another 5 minutes for testing
<reinhard> good
<johannesV> ok, my patch passed first tests successfull
<reinhard> johannesV: ok, while I'm away, please check if de translations are up to date
<johannesV> reinhard, ok
<reinhard> for at least forms, common, appserver
<reinhard> IIRC we haven't fully translated reports and designer yet anyway
<johannesV> kilo: you really get no exception whenn pressing <tab> on the qty. entry
<kilo> no
<reinhard> lol
<johannesV> ok
<johannesV> fixed that ... i don't get an exception anymore
<johannesV> :)
<reinhard> gnue-navigator had no single commit since the last release
<johannesV> kilo but everything get's saved ... even the second line
<johannesV> and i've only changed one line in INV_Head.gfd
<kilo> which one?
<reinhard> ok, prereleases done
<reinhard> bbl, off to customer
<johannesV> ok
<johannesV> hmm
<johannesV>         <trigger type="POST-CHANGE" name="onchange">
<johannesV>           #Calling a calculated property will cause it to refresh
<johannesV>           # blkINV_Item.call("INV_getvalueGross", {})
<johannesV>           blkINV_Head.call("INV_getvalueTotalGross", {})
<johannesV>         </trigger>
<johannesV> it's the call of INV_getvalueGross
<johannesV> this is a calculated field ... why don't you bind it using a field in blkINV_Item ?
<johannesV> oh
<johannesV> you *do* have such a field
<kilo> it uses find() to collect invoice lines that connect to the same invoice
<kilo> find() does not search dirty-state instances... so it does not work
<johannesV> hmm
<johannesV> but if i add a debug print to the procedure i get a line one per item-line i've added in a non-commited invoice
<kilo> hmmm
<johannesV> but there seem's something weird wrt tab-order/detail-records
<johannesV> s/seem's/seems/
<kilo> ?
<johannesV> well in the detail-block it's a bit confusing ...
<johannesV> when i'm in the note-entry, pressing tab moves to the next row's note-entry ...
<kilo> tab? it works ok at me
<kilo> hehe, now i get the exception 8-))
<johannesV> have you commented the getvalueGross-call ?
<kilo> no
<kilo> should i?
<johannesV> after i did it, the exception disappeared. anyway you won't have to do it that compicated
<johannesV> doing something like x = blkINV_Item.fldInvValueGross.get () does the same
<johannesV> and no exception occurs
<johannesV> i think the blkINV_Item has no record at that time or at least it has no 'non-empty' record
<kilo> so if i use the call(...) it raises an exception, but with .get() it does not?
<johannesV> (there's a bug in common's Base/ResultSet.post () code where an 'empty' record get's eliminated from the cache)
*** holycow has quit IRC
<johannesV> this is a problem for records which has been initialized by appserver (since they're not *dirty* and seem to be empty)
*** holycow has joined #gnuenterprise
<johannesV> at least for me the exception dissapeared ...
<johannesV> the difference between call () and get () is who does call the proc
<johannesV> on the call () method it's the *blkINV_Item*'s current record
<kilo> ah, i see
<johannesV> using the get () method it's just accessing a property (which has been filled by appserver)
<johannesV> but
<johannesV> we would have to investigate it in more detail ... to catch the *real* problem
<kilo> real problem? computers! 8-))
<johannesV> :)
<kilo> ok, now using the .get() thing, original probelm comes back again. first line dissappears after leaving the Note entry...
<kilo> calcDueDate trigger ic enabled...
*** johannesV has quit IRC
*** johannesV has joined #gnuenterprise
<kilo> have to run, bbl
*** kilo has quit IRC
*** jamest has joined #gnuenterprise
*** chillywilly has quit IRC
*** chillywilly has joined #gnuenterprise
*** johannesV_ has joined #gnuenterprise
*** chillywilly has quit IRC
*** chillywilly has joined #gnuenterprise
*** johannesV has quit IRC
*** holycow has quit IRC
<lekma> johannesV_: if i were to reload a class in a runnig server would the following be sufficient?
<lekma> def reloadClass (self, classname):
<lekma>   classdef = self.__getClassdef (classname)
<lekma>   classdef.replaceReferences (classdef.module)
<lekma>   classdef.complete ()
<lekma> the idea is only to reload
<lekma> procedures in a running server
<johannesV_> hmm
<johannesV_> what does self.__getClassdef () do ?
<lekma> it's from geasSessionManager.py, it returns a class definition
<lekma> sorry from geasSession.py
<johannesV_> replaceReferences () is used to replace instance-references loaded from repository.ini with *real* language-interface object instances
<johannesV_> it does *not* load anything new
<johannesV_> we could do something different (at least we could try)
<johannesV_> if we remove the class from sm.classes and then access it again
<lekma> from the docstring in classrep.Class
<lekma>     """
<lekma>     This class updates all references of a class to it's module object and
<lekma>     requests all properties and procedures to do update their references too.
<lekma>     """
<johannesV_> this would result in basedictionary.__getitem__ to reload the class def via find ()
<johannesV_> well, this means "references" like a unicode-string holding a gnue-id get's replaced by the real object it referres to
<johannesV_> as well as all props
<johannesV_> have a look at src/classrep/__init__.py
<johannesV_> there the base classrep is loaded from a ini-file and wrapped by Module-Dicts/Class-Dicts/Property-Dicts/...
<johannesV_> but still all refs are only strings loaded from the ini-file
*** lxf has joined #gnuenterprise
<johannesV_> this is where replaceRefernences () get's into the game
<lekma> ok
<johannesV_> it replaces all refs by a business object instance representing that reference
<johannesV_> so after the first recursive iteration of replaceReferences all following calls won't do anything
<johannesV_> since there are no strings to be replaced
<johannesV_> but let me try something other ...
<johannesV_> ok
<johannesV_> i have made a way to do it
<johannesV_> it's quite easy
<johannesV_> I've added a new method to geasSessionManager.py called "reloadClass (classname)" as well as to grpc/appserver.grpc
<johannesV_> this method looks like:
<johannesV_> def reloadClass (self, classname):
<johannesV_>     del self.classes [classname]
<johannesV_>     cDef = self.classes [classname]
<johannesV_> that's all about
<johannesV_> now one can call the session manager to reload a given class via rpc
<johannesV_> hmm
<johannesV_> but let me test a few minutes more ...
<johannesV_> ok, there's still a problem with the session-cache
<johannesV_> since this cache (internal session) still holds the *old* procedure code the method shown does not work 100%
* johannesV_ still at work ... :)
<lekma> well if you commit it works...
<lekma> ie Session.commit()
*** lxf has quit IRC
<johannesV_> ok, got it working now ...
<johannesV_> still testing ...
<johannesV_> ok, it works without a commit ()
<johannesV_> yeah, looks good
<johannesV_> it works as described above (i had a bug in the test-code first so i though cache is an issue)
<johannesV_> s/though/thought/
<johannesV_> i'll talk to reinhard if we do commit this patch into this release (since a prerelease started today)
<johannesV_> next thing to do would be to make gnue-readgcd to update a running application-server's classes ... :)
<lekma> do you mean the previous code is ok ?
<lekma> <johannesV_> def reloadClass (self, classname):
<lekma> <johannesV_>     del self.classes [classname]
<lekma> <johannesV_>     cDef = self.classes [classname]
<johannesV_> yes it's ok
<johannesV_> you don't really need the assignment to cDef, just "self.classes [classname]" would do the job too
<lekma> johannesV_: thx i'll try it before tonight
<johannesV_> classrep.Base.BaseDictionary implements __getitem__ () which does the actual reloading if the requested key is missing in the dict
<johannesV_> but
<johannesV_> we would have to add a lot of other stuff regarding server-administration as well
<johannesV_> like 'session-management'
<johannesV_> reloading parts of the classrep (like modules, classes, ...)
<johannesV_> and much more ... :)
<johannesV_> so i don't know if we have such things in svn in the near future
<lekma> :)
<lekma> the problem is that now we are a few 'functionnal developper' modifying procedures on the same server
<lekma> and always restarting the complete server
<lekma> when you want to test a modif
<lekma> is annoying to the others
<johannesV_> that's right
<johannesV_> i cannot promise to introduce this method into the current version right now, but meanwhile you could work with the patch described before ...
*** kilo has joined #gnuenterprise
*** kilo has quit IRC
*** kilo has joined #gnuenterprise
*** jamest has left #gnuenterprise
<kilo> jcater: http://www.gnuenterprise.org/gallery/forms.php, last picture. Lovely goat, ain't it?
<kilo> @weather LHUD
<bigbrother> kilo: The current temperature in Szeged, Hungary is 40°F. Conditions: Partly Cloudy. Humidity: 42%. Dew Point: 25°F. Pressure: 30.23 in (1024 hPa). Visibility: 25.0 miles (40.0 kilometers). UV: 0 out of 12
*** holycow has joined #gnuenterprise
* reinhard is confused
<reinhard> didn't we actually want the procedures to not get cached?
<reinhard> i thought that works
<reinhard> but then again
<reinhard> i guess it's better when procedures *are* cached
<lekma> well now it seems they are
<lekma> cause i have to restart to reload a proc
<reinhard> lekma: can you live with adding the lines johannesV_ gave you yourself as a temporary solution?
<lekma> yes
<reinhard> ok
<johannesV_> reinhard, still there ?
*** jamest has joined #gnuenterprise
*** sjc has joined #gnuenterprise
<lekma> johannesV_: would it be possible to do the same reloadClass with a gnue_id as parameter instead of a class name string?
<johannesV_> of course, but the procedure would change a bit ..
<johannesV_> let me try it first ...
<reinhard> johannesV_: still here
<johannesV_> lekma, here we go:
<johannesV_>   # ---------------------------------------------------------------------------
<johannesV_>   # Refresh a class definition in the class repository
<johannesV_>   # ---------------------------------------------------------------------------
<johannesV_>   def reloadClass (self, item):
<johannesV_>     """
<johannesV_>     This function refreshes a class definition in the class repository by
<johannesV_>     completely reloading it from the backend.
<johannesV_>     @param item: name or id of the class to be reloaded
<johannesV_>     """
<johannesV_>     # if item is not a classname we assume it's a gnue_id of the class
<johannesV_>     if not self.classes.has_key (item):
<johannesV_>       item = self.classes.find (item).fullName
<johannesV_>     del self.classes [item]
<johannesV_>     self.classes [item]
<lekma> thousands thanks
<johannesV_> this way you're free to use either a classname or a gnue_id
<johannesV_> oth we could screw this up a bit and add a check for module too ...
<johannesV_> like if self.modules.has_key (item) ...
<johannesV_> to add reload-caps for whole modules ...
<johannesV_> :)
<johannesV_> reinhard, i thought you were off for today ...
<johannesV_> i assume you don't like to add a feature for reloading classes/modules like this in the current state of appserver, do you ?
<reinhard> johannesV_: you assume correct
<reinhard> that's why i did
<reinhard> reinhard: lekma: can you live with adding the lines johannesV_ gave you yourself as a temporary solution?
<reinhard> lekma: yes
<reinhard> :)
<johannesV_> ok
<johannesV_> so i'm really off for today ...
<johannesV_> have a nice evening :)
*** johannesV_ has quit IRC
<lekma> johannesV_: it works perfectly
<lekma> thank you very much
<lekma> i have to leave for today (one has to eat, sleep, ...sometimes)
<lekma> :)
<lekma> thanks again
<lekma> night all
*** lekma has quit IRC
<reinhard> night all
*** reinhard has quit IRC
*** bluesbaron_ has joined #gnuenterprise
*** bluesbaron_ has left #gnuenterprise
*** holycow has quit IRC
*** ajmitch_ has joined #gnuenterprise
<kilo> weirdo... commenting code even in irc... weird austrians 8-)))
*** jcater has quit IRC
*** holycow has joined #gnuenterprise
*** jamest has left #gnuenterprise
*** kilo has quit IRC
*** sjc has quit IRC
*** holycow has quit IRC
*** holycow has joined #gnuenterprise
*** holycow has quit IRC
*** dimas has quit IRC
*** dimas has joined #gnuenterprise
*** ajmitch_ has quit IRC
*** ncjp has quit IRC
*** Vee has quit IRC
*** shmooz has quit IRC
*** havoc has quit IRC
*** agile has quit IRC
*** Morphous has quit IRC
*** involved has quit IRC
*** dimas has quit IRC
*** mustafa has quit IRC
*** nickr has quit IRC
*** chillywilly has quit IRC
*** cilkay has quit IRC
*** dneighbo has quit IRC
*** dimas has joined #gnuenterprise
*** Vee has joined #gnuenterprise
*** ncjp has joined #gnuenterprise
*** shmooz has joined #gnuenterprise
*** mustafa has joined #gnuenterprise
*** nickr has joined #gnuenterprise
*** havoc has joined #gnuenterprise
*** Morphous has joined #gnuenterprise
*** involved has joined #gnuenterprise
*** agile has joined #gnuenterprise
*** chillywilly has joined #gnuenterprise
*** cilkay has joined #gnuenterprise
*** dneighbo has joined #gnuenterprise