| Re: callable IDL and structures [message #30074] |
Tue, 09 April 2002 01:00  |
Sebastian Moeller
Messages: 7
Registered: April 2002
|
Junior Member |
|
|
Hello Mr. Fanning
this really is a nice and impressivly quick news group. Thank you very
much for answering.
David Fanning wrote:
[schnipp]
Yes, this is correct (on both counts). Ten minutes of looking through
on-line help doesn't come up with anything, but I know
it is documented *somewhere*. This is an alternative way to
access the fields of a structure. But if they are available
this way, they are sure as heck available via the fields names
as well. I'd like to see the code you are using to "access" them
by their names.
[schnapp]
OK, here we go... This is the part in the C++ project. It is supposed to
define the named structure str
typedef
struct
{
int testarr[50];
int field2;
} teststr;
IDL_MEMINT d1[10];
d1[0]=1;
d1[1]=50;
IDL_STRUCT_TAG_DEF sd[4];
sd[0].name="testarr";
sd[0].type=(void*)IDL_TYP_LONG;
sd[0].dims=(IDL_MEMINT*)d1;
sd[0].flags=NULL;
sd[1].name="field2";
sd[1].type=(void*)IDL_TYP_LONG;
sd[1].dims=NULL;
sd[1].flags=NULL;
sd[2].name=NULL;
sd[2].type=NULL;
sd[2].dims=NULL;
sd[2].flags=NULL;
IDL_MEMINT d[3];
d[0]=1;
for(int k=0; k<50; k++)
{
teststr.testarr[k]=k;
}
teststr.field2=10;
void* ss=IDL_MakeStruct("str",sd);
var=IDL_ImportNamedArray("teststr",1,d,IDL_TYP_STRUCT,(UCHAR*)(&teststr),0,ss);
int err=IDL_ExecuteStr("help, /structure, teststr");
/* this line gives the whole information for the structure, including
the tag names and tag types as defined in this c fragment (teststarr and
field2). Also the data contained in the structure (the array containing
the subscript values and 10 for field2) is shown correctly.*/
err=IDL_ExecuteStr("print, teststr.field2");
/*this line gives an error message about field2 not beeing defined in
teststr, even thoug the preceedind command showed that teststr contained
the tags testarr and field2*/
err=IDL_ExecuteStr("print, teststr.(where(TAG_NAMES(teststr) EQ
'testarr'))");
/*this line actually works, it prints the content of the array (numbers
from 0 to 49). It is a rude attempt to bring the structure tag names
back into the game, and a pretty obfusceted way to write "print,
teststr.(0)". A few defines along the lines of: #define testarr (0),
would save the day...*/
To sum it up we are able to pass the data to IDL. And IDL has all
the meta-information about the structure's contents (tags, types and
data). The tag names can even be seen from the TAG_NAMES function, but
attempts to address the tags directly fail.
I hope this information serves to illustrate the problem more deeply.
Ahoi
Sebastian Moeller
P.S.: Thios is my second attempt to post the reply, as the first seems
to be lost somewhere...
|
|
|
|
| Re: callable IDL and structures [message #30076 is a reply to message #30074] |
Mon, 08 April 2002 13:13  |
Sebastian Moeller
Messages: 7
Registered: April 2002
|
Junior Member |
|
|
Hello Mr. Fanning
this really is a nice and impressivly quick news group. Thank you very
much for answering.
David Fanning wrote:
[schnipp]
> Yes, this is correct (on both counts). Ten minutes of looking
> through on-line help doesn't come up with anything, but I know
> it is documented *somewhere*. This is an alternative way to
> access the fields of a structure. But if they are available
> this way, they are sure as heck available via the fields names
> as well. I'd like to see the code you are using to "access" them
> by their names.
[schnapp]
OK, here we go... This is the part in the C++ project. It is supposed to
define the named structure str
typedef
struct
{
int testarr[50];
int field2;
} teststr;
IDL_MEMINT d1[10];
d1[0]=1;
d1[1]=50;
IDL_STRUCT_TAG_DEF sd[4];
sd[0].name="testarr";
sd[0].type=(void*)IDL_TYP_LONG;
sd[0].dims=(IDL_MEMINT*)d1;
sd[0].flags=NULL;
sd[1].name="field2";
sd[1].type=(void*)IDL_TYP_LONG;
sd[1].dims=NULL;
sd[1].flags=NULL;
sd[2].name=NULL;
sd[2].type=NULL;
sd[2].dims=NULL;
sd[2].flags=NULL;
IDL_MEMINT d[3];
d[0]=1;
for(int k=0; k<50; k++)
{
teststr.testarr[k]=k;
}
teststr.field2=10;
void* ss=IDL_MakeStruct("str",sd);
var=IDL_ImportNamedArray("teststr",1,d,IDL_TYP_STRUCT,(UCHAR*)(&teststr),0,ss);
int err=IDL_ExecuteStr("help, /structure, teststr");
/* this line gives the whole information for the structure, including
the tag names and tag types as defined in this c fragment (teststarr and
field2). Also the data contained in the structure (the array containing
the subscript values and 10 for field2) is shown correctly.*/
err=IDL_ExecuteStr("print, teststr.field2");
/*this line gives an error message about field2 not beeing defined in
teststr, even thoug the preceedind command showed that teststr contained
the tags testarr and field2*/
err=IDL_ExecuteStr("print, teststr.(where(TAG_NAMES(teststr) EQ 'testarr'))");
/*this line actually works, it prints the content of the array (numbers
from 0 to 49). It is a rude attempt to bring the structure tag names
back into the game, and a pretty obfusceted way to write "print,
teststr.(0)". A few defines along the lines of: #define testarr (0),
would save the day...*/
To sum it up we are able to pass the data to IDL. And IDL has all the
meta-information about the structure's contents (tags, types and data).
The tag names can even be seen from the TAG_NAMES function, but attempts
to address the tags directly fail.
I hope this information serves to illustrate the problem more deeply.
Ahoi
Sebastian Moeller
|
|
|
|
| Re: callable IDL and structures [message #30082 is a reply to message #30076] |
Tue, 09 April 2002 08:33 |
Stein Vidar Hagfors H[1]
Messages: 56
Registered: February 2000
|
Member |
|
|
Sebastian Moeller <sebastian.moeller@lur.rwth-aachen.de> writes:
> Hi there IDL-experts,
>
> at my place of work we have a C++ project which is to use IDL for user
> definable output. As we are somewhat bound to windows (but do not
> really like activeX) we decided that callable IDL might be the way to
> go. We intended to use structures to pass data to IDL. The strange
> thing is now, the structures we pass show their tags if we invoke
> "help, MyStructure ,/STRUCTURE". We see all tag names and types and
> content as defined in the C++ project. But every command that tries to
> access the data in the structure from IDL (e.g. tmp=
> MyStructure.MyTag1 with MyTag1 is the first element of the structure)
> just tells that MyTag1 is not defined in MyStructure.
Just a hunch, based on your C++ background & writing things like
"MyStructure.MyTag1" etc: Try all upper (or all lower) case for the
structure tag names when *defining* them. My suspicion is that they're
converted to one of those in the "parsing" part of IDL, but since your
C++ code presumably short-circuits that and defines the structures by
calling the appropriate function directly, this might go wrong.
--
------------------------------------------------------------ --------------
Stein Vidar Hagfors Haugan
ESA SOHO SOC/European Space Agency Science Operations Coordinator for SOHO
NASA Goddard Space Flight Center, Email: shaugan@esa.nascom.nasa.gov
Mail Code 682.3, Bld. 26, Room G-1, Tel.: 1-301-286-9028/240-354-6066
Greenbelt, Maryland 20771, USA. Fax: 1-301-286-0264
------------------------------------------------------------ --------------
|
|
|
|
|
|
| Re: callable IDL and structures [message #30107 is a reply to message #30076] |
Mon, 08 April 2002 11:42 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Sebastian Moeller (sebastian.moeller@lur.rwth-aachen.de) writes:
> at my place of work we have a C++ project which is to use IDL for user
> definable output. As we are somewhat bound to windows (but do not really
> like activeX) we decided that callable IDL might be the way to go. We
> intended to use structures to pass data to IDL. The strange thing is
> now, the structures we pass show their tags if we invoke "help,
> MyStructure ,/STRUCTURE". We see all tag names and types and content as
> defined in the C++ project. But every command that tries to access the
> data in the structure from IDL (e.g. tmp= MyStructure.MyTag1 with MyTag1
> is the first element of the structure) just tells that MyTag1 is not
> defined in MyStructure.
I'm afraid I'm skeptical about this. :-)
> Beeing challenged in that way we found that tmp= MyStructure.(n) with n
> beeing the "address" of the tag, actually gives us access to the data.
> "MyStructure.[0]" by the way does not work.
Yes, this is correct (on both counts). Ten minutes of looking
through on-line help doesn't come up with anything, but I know
it is documented *somewhere*. This is an alternative way to
access the fields of a structure. But if they are available
this way, they are sure as heck available via the fields names
as well. I'd like to see the code you are using to "access" them
by their names.
> But as this behavior is not
> documented anywhere (well, at least not where we searched) we assume it
> is rather daring to go relay on the stability of this feature for the
> future. (Heck, RSI even changed the behavior of SIZE() between 5.3 and
> 5.4, without mentioning in the what's new in 5.4. So maybe stability is
> relative anyhow ;)...)
I think you are fine doing this.
> So after too long a story, is there anybody out there who knows where we
> went wrong or whether this might be considered a bug worth reporting to RSI?
I don't see any bugs...yet.
Cheers,
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: callable IDL and structures [message #30187 is a reply to message #30082] |
Thu, 11 April 2002 04:25 |
Sebastian Moeller
Messages: 7
Registered: April 2002
|
Junior Member |
|
|
Hi
Thank you very much David, wayne and Stein. All of your comments have
been very helpful. It took some time though to get back to the C++
machine but now things work likke they are supposed to be. While it was
assuring to find that structure.(tag_address) actually is documented,
what really saved the day is the fact that the tag names had to be
declarded ALL UPPERCASE in the C++ project. Since we done that things
work fine.
Ahoi & thanks a load
Sebastian Moeller
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
