bhallalab / moose Goto Github PK

Perhaps just define a global or a function isMOOSE

Reported by: upibhalla

Indexing of element⁰ should be equivalent to ‘element’. Currently we have

create neutral foo⁰

showfield foo⁰ * // works
showfield foo * // fails

Also there seem to be issues when foo⁰ is created using a simundump.

Reported by: upibhalla

In the case of the Kholodenko regression test, neardiff lets the comparison through. However, a graphical plot reveals that the curves differ widely.

Reported by: niraj_

If “step” is issued without arguments, the scheduler advances time by either 1 or 2 steps.

Can’t think of a simple script to reproduce this problem (“getstat” is not working).

This problem was discovered in a script that looked like this:

for ( i = 0; i < 3000; i = i + 1 )
step
end

A plot object was connected to a compartment, and stored around 4500 points in the output file. In particular, the very first call of “step” steps twice. Subsequently, the system alternates between behaving and misbehaving according to some pattern (did not investigate).

Should also check if “step” behaves well for all integer arguments.

Reported by: niraj_

WildCard currently does not handle regular expressions. It should handle them.

Reported by: raameshd

create compartment c
create compartment c
<Error: Neutral::create: Attempt to overwrite existing element ‘/c’. Using original.>
setfield ^ x 10
<Segmentation Fault>

Reported by: raameshd

The ‘exists’ command should have the syntax
exists element [field]

Currently it ignores the field option

Reported by: upibhalla

create Compartment c
addmsg c/channel h/channel
<Error in doAdd c1/channel h/channel>
Instead it should print <unknown object h> or <h not found> something like that.

Reported by: raameshd

Assertion fails on doing ‘showmsg’ for any element.

Reported by: niraj_

Field access in objects named Xa, where X can be any char, do not work except in root. This causes problems in /library/Na and /library/Ca. Checked in as TESTS/bugs/naming_Xa.g.

Reported by: raameshd

The MOOSE SLI command-line interface does not handle all the nice features of the GENESIS command-line interface, like control-P for previous command and in-line text editing. This is a hard one to fix because it would involve implementing a lot of operations, and also we would need ncurses or something similar, which is not terribly portable.

Reported by: upibhalla

When we do
ce noname
when there is no object name noname we should get an error, but we don’t get.

Reported by: raameshd

the default simpath in linux is current working directory ( . ) and user’s home directory. But it remains empty in windows.

Reported by: subhacom

When the file pulse.plot already exists,

tab2file pulse.plot tab table -nentries 199 -overwrite

actually appends the contents of table ‘tab’ to the file instead of overwriting.

Reported by: subhacom

In legacy kkit models, it is possible to send a SLAVE message to a molecule with a concentration value. This requires the molecule to convert it to ‘n’ values. MOOSE uses SUMTOTALS to handle slave messages, but in this case fails to do the conversion from conc to ‘n’. This problem affects Molecule.cpp as well as KineticHub.cpp.

Likely to be rather rare, only used in kkit.

Reported by: upibhalla

The MOOSE interpreter (using the GENESIS language interface) does not farm out unrecognized commands to the UNIX shell. This is a nice convenience but would take a little work to behave on all platforms.

Reported by: upibhalla

The SIMPATH variable in GENESIS is used to tell the system which directories to search in. MOOSE does not handle this, and as a result scripts with default directories (like kkit dependent scripts) do not work unless the required files are in the working directory.

Reported by: upibhalla

The command:
copy /library/cc /cell/cc
copies cc under root element.

However:
ce /
copy /library/cc cell/cc
works.

File: copy_absolute.g

Reported by: niraj_

In the MOOSE build tree, we have the utility directory, and then within it the randnum directory. This organization messes up the compilation unless the randnum directory is last. This is because the loop for building subdirectories assumes that all are just under the main MOOSE code directory. The randnum directory should be brought down to the same level as the others.

Reported by: upibhalla

The following lines appear on stderr on startup:
ERROR: failed to open property file: config.xml
ERROR: PropertyReader::getProperty( AutoSchedule ) :: No such key exists in property map.
ERROR: PropertyReader::getProperty( CreateSolver ) :: No such key exists in property map.

Should be handled quietly or go to stdout at most.

Reported by: upibhalla

On stepping trough the execution it was found that vector< Conn >::const_iterator SimpleElement::connDestEnd( unsigned int dest ) const

returns an iterator which is not dereferancable. In method void DestFinfo::dropAll( Element* e ) const this iterator is being dereference and this causes an assertion failure, causing program termination.
Below is the call stack in Visual Studio 2005 version 8.0.50727.42 with Microsoft .NET Framework version 2.0.50727:

> moose.exe!DestFinfo::dropAll(Element * e=0×00566918) Line 65 C++
moose.exe!ClockJob::clearMessages(Element * e=0×00566918) Line 408 + 0×2f bytes C++
moose.exe!ClockJob::reschedFuncLocal(Element * e=0×005666d8) Line 369 + 0×21 bytes C++
moose.exe!ClockJob::reschedFunc(const Conn & c={…}) Line 320 C++
moose.exe!set(Element * e=0×005666d8, const Finfo * f=0×00556d60) Line 30 + 0×13 bytes C++
moose.exe!set(Element * e=0×005666d8, const std::basic_string<char,std::char_traits<char>,std::allocator<char> > & f=“resched”) Line 45 + 0×7 bytes C++
moose.exe!testSchedProcess() Line 334 + 0×3f bytes C++
msvcp80.dll!7c44f3e5()
[Frames below may be incorrect and/or missing, no symbols loaded for msvcp80.dll]
moose.exe!set<int>(Element * e=0×00000000, const std::basic_string<char,std::char_traits<char>,std::allocator<char> > & f={…}, int v=-1479030319) Line 50 + 0xc bytes C++
moose.exe!testSched() Line 164 + 0×10 bytes C++
msvcr80.dll!78159301()
msvcp80.dll!7c42319e()
moose.exe!main(int argc=1, char * * argv=0×00355728) Line 88 C++
moose.exe!__tmainCRTStartup() Line 586 + 0×17 bytes C
kernel32.dll!7c816d4f()
kernel32.dll!7c8399f3()

Reported by: subhacom

SynChan lookup fields weight and delay remain inaccessible unless a reset is issued. This comes in the way of building the model.

Reported by: niraj_

Wildcards not handled by showfield

Reported by: upibhalla

Doing “showfield *” after adding a message to a field shows the field twice.

Script attached.

Reported by: niraj_

The showmsg command on a newly created object causes assertion failure. An example session:

moose #1 > create compartment /c
moose #2 > showmsg /c
INCOMING MESSAGES onto /c
moose: SimpleElement.cpp:196: virtual const Msg* SimpleElement::msg(unsigned int) const: Assertion `msgNum < msg_.size()’ failed.
Aborted (core dumped)

Reported by: subhacom

create Compartment c
setfield c x notanumber

This gives no error.

Reported by: raameshd

“sendback” seems to consider only the first message in the linked list of messages. This means a callback will always invoke the RecvFunc from a single class, even if the callback request came from an object of another class.

Might the same problem exist in “sendTo” as well?

Reported by: niraj_

In floating point arithmetic, the result calculated is sometimes lacks in precision. In the following example, the precision loss is quite severe:

moose #7> echo { 10.0*2e-5/(3.141592*1.5*1.5e-10) }
282942

whereas the correct result is: 28294.217991670746

Reported by: subhacom

> I am compiling moose on Fedora Core 8 and getting following error message
> ld: cannot find -lm
> make¹: * [randnum.o] Error 1
> make¹: leaving directory ‘/root/moose-beta-1.0.0.src/utility/randnum’
> All libs compiled
> make: * No rule to make target ‘utility/utility.o’, needed by moose.
> stop

Reported by: Shesharao M. Wanjerkhede

Reported by: subhacom

showmsg <object> shows same messages under the incoming header and the outgoing header. The listMessages function in Shell.cpp does not use the parameter bool isIncoming anymore. So it returns a list of both incoming messages and outgoing messages.

Reported by: subhacom

An object with an absolute path is created in the
current working element.
create neutral /n
ce /n
create neutral /foo
le
> foo

Reported by: *anonymous

Element stack doesn’t work. Current working element stays unchanged. Showfield leads to segmentation fault.

File: TESTS/bugs/elstack.g

Reported by: niraj_

Readcell does not honour global variables, irrespective of whether they are set in the .g or .p file.

Reported by: niraj_

MOOSE crashes if it is started up with an argument of a nonexistent script file. Also in the ‘include’ statement if a nonexistent script file name is used. Error message claims double free or corruption.

Reported by: upibhalla

Refer to TESTS/regression/moose_readcell.g

Compartment names like axon¹ are not handled properly. The reason is that ‘axon¹’ is assumed to exist because ‘axon’ does (in this particular example). Go to ReadCell.cpp:298 in revision 450.

Reported by: niraj_

These files are generated by neuroConstruct. Rm and Cm aren’t being picked up from attached *.p file. Error on readcell:

Warning: Ignored attempt to set Ra of compartment Soma to less than 1e-15
Warning: Ignored attempt to set Cm of compartment Soma to less than 1e-15

Also throws error at reset:

Error: Neutral::create: class HSolve not found
moose: Cell.cpp:272: void Cell::setupSolver(const Id&, const Id&) const: Assertion `integ != 0’ failed.
Aborted

Reported by: *anonymous

In ²: r = moose.Neutral(“/”)

In ³: for i in r.children:
…: print i.path()
…:
…:
/shell
/sched
/library
/proto
/solvers
Segmentation fault (core dumped)

Reported by: subhacom

create Compartment c
setfield c x
moose: Shell.cpp:1206: static void Shell::setVecField(const Conn&, std::vector<Id, std::allocator<Id> >, std::string, std::string): Assertion `i->good()’ failed.
Aborted (core dumped)

Reported by: raameshd

create \ <Enter>
and after that you have no option but to ^C or close the shell. Because everything after that gives error.

Reported by: raameshd

Not able to duplicate the gate message when successive copy of the channel is made. Such copying is often used. /library has prototype channels. readcell copies the prototypes in library and create the prototype neuron. When this prototype is copied using createmap or copy, channel messages are no where to be found.

Reported by: raameshd

create neutral /foo
ce /foo
create neutral bar
copy bar zod

zod will be made on /, not on /foo.

I thought I had fixed this but it has resurfaced.

It causes failure of the moose_readcell.g regression test.

Reported by: upibhalla

The command:
echo {getstat -time}
always prints “3” as a result.

In fact, giving any arbitrary arguments, or no arguments at all also behaves in the same way.

Reported by: niraj_

When ‘make pymoose’ is run, the compilation fails with:

g++ -g -DDO_UNIT_TESTS -fPIC -c moose_wrap.cxx -o moose_wrap.o -I/usr/local/include/python2.5 -I..
moose_wrap.cxx: In function ‘int SWIG_Python_ConvertFunctionPtr(PyObject*, void**, swig_type_info*)’:
moose_wrap.cxx:2052: error: invalid conversion from ‘const char*’ to ‘char*’
moose_wrap.cxx: In function ‘void SWIG_Python_FixMethods(PyMethodDef*, swig_const_info*, swig_type_info**, swig_type_info**)’:
moose_wrap.cxx:64849: error: invalid conversion from ‘const char*’ to ‘char*’
make¹: * [moose_wrap.o] Error 1

Reported by: subhacom

In certain cases, after automatic solver setup by Cell (i.e., the solver manager), ticks do not point to their intended targets. This leads to segfaults, since the RecvFunc is correct, but target element is something else entirely.

File: TESTS/bugs/solve_sched.g

Reported by: niraj_

If a cell is created using ReadCell, then the compartment-synchan message does not get added properly. Along with the usual channel message, another axial message also gets added between the two.

This will happen irrespective of whether the synchan is created using readcell, or via the shell.

File: Synchan.g in TESTS/bugs/synmsg

Reported by: niraj_

Accessing a channel field after accessing a gate field does not work. This happens only if both of the above are done in a single statement.

Example:

/* Works. */
setfield K_mit_usb Ek 0.07 Y_A>sy 5.0 Y_B->sy 5.0

/* Doesn’t work. */
setfield K_mit_usb Y_A->sy 5.0 Y_B->sy 5.0 Ek -0.07

Error message:
Error: cannot find field: B.Ek

Example script attached.

Reported by: niraj_

The getpath command needs to be implemented to manipulate path strings.

Reported by: upibhalla

ExtFieldFinfo is a new finfo to add new fields, dynamically, by addfield command. The strSet function, in ExtFieldFinfo, is const function, whose function is to set the value of a private variable. We also need to see that this added field is able to connect to other standard fields.

Currently, this has been handled by doing a const_cast.

Reported by: raameshd

This happens only when a Compartment instance has been created. The SIGSEGV occurs at Compartment::initReinitFunc(const Conn& c, ProcInfo p) – when c.data()->innerInitReinitFunc is called. c.targetElement()->name() gives the correct name of the compartment, but it is not clear why the call to innerInitReinitFunc is causing segmentation fault although c.data() is non-null.
innerInitReinitFunc is an empty function.

Reported by: subhacom

The script seems to be outdated. Might need some correction.
Error: in IsCommand: Did not find command ‘EREST_ACT’
Error: Neutral::create: Attempt to overwrite existing element ‘/library’. Using original.
soma.p read: 1 compartments, 2 channels, 0 others
Error: setfield : cannot find element /axon/soma/Na_mit_usb/xGate/A
Error: setfield : cannot find element /axon/soma/Na_mit_usb/xGate/B
Error: setfield : cannot find element /axon/soma/Na_mit_usb/yGate/A
Error: setfield : cannot find element /axon/soma/Na_mit_usb/yGate/B
Error: setfield : cannot find element /axon/soma/K_mit_usb/xGate/A
Error: setfield : cannot find element /axon/soma/K_mit_usb/xGate/B
Error: setfield : cannot find element /axon/soma/K_mit_usb/yGate/A
Error: setfield : cannot find element /axon/soma/K_mit_usb/yGate/B

Reported by: subhacom