According to the "Introduction to Uniform Type Identifiers "
document: "Support for uniform type identifiers is available in Mac
S X v10.3 and later."
According to the "System-Declared Uniform Type Identifiers: document:
Table 3-1 lists the uniform type identifiers that are declared by the
system as of Mac S X v10.4.
I recently upgraded a file importing mechanism in my program to use
the Uniform Type Identifiers scheme and its associated methods. This
works beautifully in 10.4, but users running 10.3.9 are encountering
crashes when importing files. I believe I have tracked it down to the
UTI references.
According to bit of documentation, UTIs should work in 10.3.9, but
according to another bit of documentation, the constants used by the
UTI implementation such as kUTTypeText or kUTTypeWebArchive are
available as of 10.4. What's going on here? Am I allowed to use the
UTI scheme in 10.3.9 as long as I hardcode the UTI definitions, ie
@"com.apple.webarchive" for kUTTypeWebArchive. The crash log does not
offer much in the way of usefulness when the app goes down during an
import, but if I had to guess, it would be because I'm referencing
symbols that are only included in 10.4 and above.
Can anyone confirm that hypothesis?
-Phil
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com
This email sent to bsdarchive (AT) developershed (DOT) com

1/3/06, Philip Dow <dev (AT) phildow (DOT) netwrote:
According to bit of documentation, UTIs should work in 10.3.9, but
according to another bit of documentation, the constants used by the
UTI implementation such as kUTTypeText or kUTTypeWebArchive are
available as of 10.4. What's going on here? Am I allowed to use the
UTI scheme in 10.3.9 as long as I hardcode the UTI definitions, ie
@"com.apple.webarchive" for kUTTypeWebArchive. The crash log does not
offer much in the way of usefulness when the app goes down during an
import, but if I had to guess, it would be because I'm referencing
symbols that are only included in 10.4 and above.

Can anyone confirm that hypothesis?

Cross-development support in Xcode and the SDK exist to help detect
issues like the above. (search ADC for "cross development")

Anyway looking at the header that defines "kUTTypeWebArchive"
(UTCoreType.h) I see the following

extern const CFStringRef kUTTypeWebArchive
;

So you could use cross-development and weak linking to check if that
is available at run-time and if not use a definition from your own
binary.
-Shawn

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

Jan 3, 2006, at 1:48 PM, Philip Dow wrote:

According to the "Introduction to Uniform Type Identifiers
" document: "Support for uniform type identifiers is
available in Mac S X v10.3 and later."

According to the "System-Declared Uniform Type Identifiers:
document: Table 3-1 lists the uniform type identifiers that are
declared by the system as of Mac S X v10.4.

I recently upgraded a file importing mechanism in my program to use
the Uniform Type Identifiers scheme and its associated methods.
This works beautifully in 10.4, but users running 10.3.9 are
encountering crashes when importing files. I believe I have tracked
it down to the UTI references.

According to bit of documentation, UTIs should work in 10.3.9, but
according to another bit of documentation, the constants used by
the UTI implementation such as kUTTypeText or kUTTypeWebArchive are
available as of 10.4. What's going on here? Am I allowed to use the
UTI scheme in 10.3.9 as long as I hardcode the UTI definitions, ie
@"com.apple.webarchive" for kUTTypeWebArchive. The crash log does
not offer much in the way of usefulness when the app goes down
during an import, but if I had to guess, it would be because I'm
referencing symbols that are only included in 10.4 and above.

Can anyone confirm that hypothesis?

I seem to recall that there was a recent posting, either here or on
the Carbon list, about this topic. I believe the replier said that
you would have to provide your own UTI strings for any 10.4+ UTI
strings if you're running on a pre-10.4 system. You could certainly
define a class with methods that do the right thing for both cases so
that callers don't have to know the gory details, something like:

+ (NSString*) getUTTypeWebArchive
{
#if MACS_X_VERSIN_MIN_REQUIRED < MACS_X_VERSIN_10_4
if (kUTTypeWebArchive == NULL)
return @"com.apple.webarchive";
#endif

return (NSString*) kUTTypeWebArchive;
}

steve

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

Thanks Shawn and Steve. I thought it might have to do with the web
archive uti if not the uti constants in general. I'm already weak
linking my app and checking for 10.4 before calling other methods, so
I should be able to slide this in without too much difficulty.
-Phil

Jan 3, 2006, at 11:41 PM, Steve Christensen wrote:

Jan 3, 2006, at 1:48 PM, Philip Dow wrote:
>
>According to the "Introduction to Uniform Type Identifiers
>" document: "Support for uniform type identifiers is
>available in Mac S X v10.3 and later."
>>
>According to the "System-Declared Uniform Type Identifiers:
>document: Table 3-1 lists the uniform type identifiers that are
>declared by the system as of Mac S X v10.4.
>>
>I recently upgraded a file importing mechanism in my program to
>use the Uniform Type Identifiers scheme and its associated
>methods. This works beautifully in 10.4, but users running 10.3.9
>are encountering crashes when importing files. I believe I have
>tracked it down to the UTI references.
>>
>According to bit of documentation, UTIs should work in 10.3.9, but
>according to another bit of documentation, the constants used by
>the UTI implementation such as kUTTypeText or kUTTypeWebArchive
>are available as of 10.4. What's going on here? Am I allowed to
>use the UTI scheme in 10.3.9 as long as I hardcode the UTI
>definitions, ie @"com.apple.webarchive" for kUTTypeWebArchive. The
>crash log does not offer much in the way of usefulness when the
>app goes down during an import, but if I had to guess, it would be
>because I'm referencing symbols that are only included in 10.4 and
>above.
>>
>Can anyone confirm that hypothesis?
>
I seem to recall that there was a recent posting, either here or on
the Carbon list, about this topic. I believe the replier said that
you would have to provide your own UTI strings for any 10.4+ UTI
strings if you're running on a pre-10.4 system. You could certainly
define a class with methods that do the right thing for both cases
so that callers don't have to know the gory details, something like:

+ (NSString*) getUTTypeWebArchive
{
#if MACS_X_VERSIN_MIN_REQUIRED < MACS_X_VERSIN_10_4
if (kUTTypeWebArchive == NULL)
return @"com.apple.webarchive";
#endif

return (NSString*) kUTTypeWebArchive;
}
--
steve

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40phildow.net

This email sent to dev (AT) phildow (DOT) net

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

1/3/06, Philip Dow <dev (AT) phildow (DOT) netwrote:
According to bit of documentation, UTIs should work in 10.3.9, but
according to another bit of documentation, the constants used by the
UTI implementation such as kUTTypeText or kUTTypeWebArchive are
available as of 10.4. What's going on here? Am I allowed to use the
UTI scheme in 10.3.9 as long as I hardcode the UTI definitions, ie
@"com.apple.webarchive" for kUTTypeWebArchive. The crash log does not
offer much in the way of usefulness when the app goes down during an
import, but if I had to guess, it would be because I'm referencing
symbols that are only included in 10.4 and above.

Can anyone confirm that hypothesis?

Cross-development support in Xcode and the SDK exist to help detect
issues like the above. (search ADC for "cross development")

Anyway looking at the header that defines "kUTTypeWebArchive"
(UTCoreType.h) I see the following

extern const CFStringRef kUTTypeWebArchive
;

So you could use cross-development and weak linking to check if that
is available at run-time and if not use a definition from your own
binary.
-Shawn

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40mac.com

This email sent to bluegus (AT) mac (DOT) com

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

As a follow up

I am noticing that the UTI scheme does not always identify the super
type of a file. I have users trying to import regular text files
identified by extensions such as .en but the method UTTypeConformsTo
does not return true when checking this file against
kUTTypePlainText. Is there an easy way for a user to set the file
type of a file or many files without changing the extension. In the
Finder, you can choose get info to change the associated application.
Is there something like that for file type?
-Phil

Jan 4, 2006, at 1:48 AM, August Trometer wrote:

1/3/06, Philip Dow <dev (AT) phildow (DOT) netwrote:
>According to bit of documentation, UTIs should work in 10.3.9, but
>according to another bit of documentation, the constants used by the
>UTI implementation such as kUTTypeText or kUTTypeWebArchive are
>available as of 10.4. What's going on here? Am I allowed to use the
>UTI scheme in 10.3.9 as long as I hardcode the UTI definitions, ie
>@"com.apple.webarchive" for kUTTypeWebArchive. The crash log does not
>offer much in the way of usefulness when the app goes down during an
>import, but if I had to guess, it would be because I'm referencing
>symbols that are only included in 10.4 and above.
>>
>Can anyone confirm that hypothesis?
>
Cross-development support in Xcode and the SDK exist to help detect
issues like the above. (search ADC for "cross development")
--
Anyway looking at the header that defines "kUTTypeWebArchive"
(UTCoreType.h) I see the following

extern const CFStringRef kUTTypeWebArchive
;

So you could use cross-development and weak linking to check if that
is available at run-time and if not use a definition from your own
binary.

-Shawn

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40mac.com

This email sent to bluegus (AT) mac (DOT) com

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

Jan 4, 2006, at 3:57 AM, Philip Dow wrote:

As a follow up

I am noticing that the UTI scheme does not always identify the
super type of a file. I have users trying to import regular text
files identified by extensions such as .en but the method
UTTypeConformsTo does not return true when checking this file
against kUTTypePlainText. Is there an easy way for a user to set
the file type of a file or many files without changing the extension.

I don't think there is any easy way for a *user* (without additional
tools) to do this. An application or the system has to export a UTI
which says .en is text, or the file has to have an HFS type of 'TEXT'.

Jim

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

Jan 3, 2006, at 2:44 PM, Philip Dow wrote:

Thanks Shawn and Steve. I thought it might have to do with the web
archive uti if not the uti constants in general. I'm already weak
linking my app and checking for 10.4 before calling other methods

In general you shouldn't need to check for 10.4 just if the symbol
you want to use exists (non-null).
-Shawn

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

I'm using 10.4 features all over the app within a variety of objects
while continuing compatibility with 10.3.9. To accomplish this,
rather than check for the existence of a 10.4 specific symbol
whenever I need it, I check NSAppKitVersionNumber as my main
controller inits and then set a user default bool depending on the
value. I've done that, the bool value is available from anywhere
with minimal hassle.
-Phil

Jan 4, 2006, at 12:01 AM, Shawn Erickson wrote:

Jan 3, 2006, at 2:44 PM, Philip Dow wrote:
>
>Thanks Shawn and Steve. I thought it might have to do with the web
>archive uti if not the uti constants in general. I'm already weak
>linking my app and checking for 10.4 before calling other methods
>
In general you shouldn't need to check for 10.4 just if the symbol
you want to use exists (non-null).

-Shawn

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com

Jan 4, 2006, at 8:33 AM, Philip Dow wrote:

Jan 4, 2006, at 12:01 AM, Shawn Erickson wrote:
>
>Jan 3, 2006, at 2:44 PM, Philip Dow wrote:
>>
Thanks Shawn and Steve. I thought it might have to do with the
web archive uti if not the uti constants in general. I'm already
weak linking my app and checking for 10.4 before calling other
methods
>>
>In general you shouldn't need to check for 10.4 just if the symbol
>you want to use exists (non-null).
>
I'm using 10.4 features all over the app within a variety of
objects while continuing compatibility with 10.3.9. To accomplish
this, rather than check for the existence of a 10.4 specific symbol
whenever I need it, I check NSAppKitVersionNumber as my main
controller inits and then set a user default bool depending on the
value. I've done that, the bool value is available from
anywhere with minimal hassle.

You're having to do a runtime check in either case so having a
special boolean doesn't really buy you anything. With a boolean (or
even saving the value returned by NSAppKitVersionNumber), you have to
keep track of the first S version where a symbol becomes available;
checking for the existence of a particular symbol lets you
automatically handle the case where, say, a symbol becomes available
in 10.4.3 but wasn't available in 10.4.

Also, wrapping these specifics in a few class methods or routines
means that the S version details are also limited to a few specific
places, rather than scattered all over your code. Bug fixes happen in
the one instance of a wrapper and are automatically available to all
callers.

steve

Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (Cocoa-dev (AT) lists (DOT) apple.com)
Help/Unsubscribe/Update your Subscription:
%40developershed.com

This email sent to bsdarchive (AT) developershed (DOT) com