If you insist to *buy* e.g. for any esoteric features of Enterprise editions, Microsoft used to have a downgrade policy so you'd buy the latest, then receive the older media and keys. Just ask their license hotline before you shell out any money, this policy was limited to the US and not available in Germany. Here instead they gave straight recommendations for companies that specialize in oldtimer software - of course by then I already had the newer package. Dirk ... View more

Try something like http://www.microsoft.com/downloads/details.aspx?FamilyId=3254C868-BCB9-412C-95C6-D100C872EC60&displaylang=en Dirk ... View more

The startup script loader for the built-in scripts such as XHTML now uses app.doScript, but I think a few versions ago it took $.evalFile and it should still work. Dirk ... View more

You can omit the trailing ";" on the #include . There are more alternatives. #include is best if you want to combine multiple script files during startup of a lengthy script. The builtin method eval() can work within a scope - e.g. when you invoke it within a function call, it can see and modify locals. Occasionally it is faster to eval a string than to load it from disk. $.evalFile works like eval(), but given a file. app.DoScript has additional parameters to control InDesign undo behaviour. Dirk ... View more

Do you have an older copy of the plugin around? Also see other paths as specified via plugin config. Then re-check the IDs, but this time for mismatches in the IDspace constant! Dirk ... View more

1) yup. Understandable, though, as IAppearancePlaceBehaviorUI is only referenced in one boss in the same plugin, and as the same boss also has a private implementation kAppearanceHierarchyNodeImpl for IHierarchy, someone might have taken shortcuts here. 2) would be never reached if 1) already crashes. 3) Probably (I did not check inheritance of AppearancePlaceBehaviorUI ) the whole thing just walks back to sourceItem (unless *this is a second child of its parent hierarchy) and could be optimized away. Ah, it is also a weird way to say const_cast ! Probably a mindless copy-paste job from somewhere else, then incompletely edited. 4) yup, see 3). 5) yup. My bottom line: I still prefer to see more real world source code that explains a concept, rather than have everything hidden in binary until it is polished to highest standards. Do you know the bug report page? Sometimes suggestions make it thru. Regarding true exceptions, these have been a big no-go until including CS4. I think CS5 finally enables them by default for plugin projects, but I would not throw across stackframes of foreign / InDesign's own code, or even more general Adobe frameworks. OT: Given your subject line I had hoped for a discussion on the real error handling via global ErrorCode and alike, but unfortunately such a discussion would anyway be very short. We had it during this year's summit - you're not supposed to set the global errorcode unless you're a command. Especially not if you're an observer, most code that executes commands is not prepared for arbitrary failure. Try it yourself to see neat "Protective Shutdown" dialogs. Dirk ... View more

Checking for isAttached hides ASSERTs that otherwise would be an indicator for unbalanced or duplicate Attach/Detach. I prefer to fix the code rather than suppress the error message. Dirk ... View more

Action components are only responsible for (menu and keyboard-shortcut) actions. To understand the selection subsystem, refer to the programming guide and SDK examples, good search keywords are Suite and CSB . Dirk ... View more

Some more ideas below. Of course any of them could also make it worse. Experimentation will show, I just don't know. Do you invoke the process from outside (e.g. SOAP) per document? Try to pile up a list of multiple jobs and process them from one run. Of course you already know about those additional parameters to app.doScript - see the OMV section of ESTK help on UndoModes for details. If the number of templates is limited and your modifications are not too destructive, you could keep the document open for reuse, and just fill in other XML. Alternatively try a full undo, or revert to saved, rather than instantiating templates. Dependent on the amount of data, try the ExtendScript XML object rather than XMLElement / import. If your script can run in a permanent target engine (other than "main"), you need not load it every time. Be careful with memory leaks, though. How complicated is the static/unchanged part of the template? Maybe that part of the document is faster as placed and eventually even embedded PDF, or EPS. Are there images that you could pre-scale and crop once rather than let InDesign do it? Also revisit their color spaces to match your PDF export settings. Minimize your template, strip all unused fonts, styles, swatches. Watch out for the flattener (output preview). On the Mac, use Instruments from the Developer tools, and there the Sampler. It allows to watch the server while it is running. That should give some statistics where it is really spending most time. Use $.hiresTimer for measurements. Dirk ... View more

The constraint to use only one .fr file is real but a very annoyance. IMHO it must've taken more time to cripple the odfrc compiler down to that requirement (the original Apple Rez compiler did not have such limitations) than it would have taken to make it going. Anyway, I usually have some postprocess script that will merge multiple resource sets, unfortunately with the new resource architecture that will be a rewrite. Maybe use tar? I haven't had the requirement for CS5 yet, but some bigger plugins already look around the corner. There plenty panels and dialogs heavily reuse the same conditionally built macros, so I can't just chain load them into one umbrella fr - even if I had the patience to wait for the compiler. Just my 0.02 EUR Dirk ... View more

You're creating a sub-menu on the file menu where you just want a menu item. Does your kPSUIPluginsMenuPath eventually have a final colon? It should be  "Main:&File"  . Dirk ... View more

Could it be that you have the release build running and try to launch the debug build, or vice versa? Btw, you can also attach the debugger  to a running program, after launch. Menu / Run / Attach to Process ... Dirk ... View more

With FindWidget it is the other way around - it is equivalent to Get() rather than to Query(). So you'd assign its result to an IControlView*  cv, or add IID_ICONTROLVIEW to the InterfacePtr constructor. It takes a while to get used to this reference counting. Dirk ... View more

You're introducing a boss leak. QueryPanelByWidgetID yields a reference counted value, you must store it within an InterfacePtr<> or explicitly invoke Release(). Besides those wide multi-step expressions are bad to read, even with nowadays monitors, and debug (set a breakpoint?). Finally defensive code should check pointers, e.g. before you invoke SetString(). Sometime later you might want to modify the underlying resource to use a different widget type, remove the widget for localization or make it optional... Also have a look at the header comments, QueryPanelByWidgetID may yield NULL if the panel is not showing. Dirk ... View more

These files are created by the ODFRC compiler. (gives a great search keyword in the docs) Make sure that in VC the path to that external compiler is configured to point into the correct SDK folder. Also compare to the file properties or the underlying XML of .fr sources in working example projects, you might have forgotten to invoke the build step in yours. Dirk ... View more

Yes. See the top of the scripting forum, this is SDK forum about C++ plugin development. Dirk ... View more

No experiences here with that exact combination. Something general: Make sure access rights on the application support folder are correctly set for the user account that you're using to launch InDesign. Run the OSX repair privileges utility. Compare the general folder structure to your desktop installation, or the earlier installation (replacing an old hardware?). If that fails: Launch the application from within Instruments (which comes with the optional installation of developer tools, e.g. XCode) Trace file access and watch out for error results. Dirk ... View more

Sorry, I guess most here are either busy with their own work or on vacation. Or both ... Besides, the focus of this forum is C++ plugin development. Java questions are better adressed at the next door "Scripting", because Java goes thru the same high level API as all the other language bindings. Rephrasing the problem with JavaScript (should be a no-brainer) will give you the broadest audience there. The Server forum would also be an alternative, especially if the problem is really Server / Java specific. Dirk ... View more

The easiest way is to step backward in version control until the crash disappears, or to selectively turn off features by compile flag. If the problem goes away, you know where to look. Beyond that, I'd first try to reproduce the crash on a Mac, because crash dumps are by far better readable there. If that fails, I'd start with the storage location of that DEADBEEF constant, somewhere within a larger block. Using breakpoints in assembly code, I'd follow the code backward to see how the memory is reached. If it is the same location after restart, I'd also try watchpoints to see earlier valid data. Of course details of the observation all depend on: - whether the crashing code is mine, InDesign's, or within the OS - how execution gets there - which allocation routines are responsible for the memory block in question - familiarity with the respective memory management (InDesign boss, InDesign other, external library, OS, your own ...) - familiarity with the required tools (debugger, compiler, understanding of the generated code at assembler level) and so forth. In general such investigations are far beyond the scope of a forum - you can easily waste a week or more just stepping around in assembler. Dirk ... View more

0xDEADBEEF is used as fill word for free'd memory. When reported as location, that means you got a pointer from such un-allocated memory. Dirk ... View more

The reference cogs (those without the pen) were fixated on the paper with needles / pins, of course that required a layer of cardboard below the paper. There were multiple holes in the cogs with different distance from the center, so you could get thru with very few iterations (e.g. even 3 spins) per path, then slightly vary the distance and color and restart. Another possibility to vary here - you'd offset the pen cog by one step after completion of a path, and eventually combine that with different distances. Aside from your ellipsis distortion, advanced patterns could also be done with multiple reference cogs positioned next to each other so your pen cog could run around the whole group of them. There were also a few different shapes - at least one straight line (of course with round caps) and a ring where the drawing cog was moving inside. If you were careful you could even use the ring for drawing, but it would just too easily jump away from the pinned cog. Dirk ... View more

Only the plugin developer can add scripting support to the plugin. See the plugin's documentation or browse the generated object model using ESTK to find such added functionality. Dirk ... View more

Thanks for the detailed report. Have there been any similar validation efforts with "VS 2008 Express Editions with SP1" ? The free download is still available here: http://www.microsoft.com/downloads/details.aspx?FamilyId=3254C868-BCB9-412C-95C6-D100C872EC60&displaylang=en Dirk ... View more

I haven't done such tricks for ages, but I would use a regular control/widget as base implementation, overloading CControlView:: GetSysControl or GetSysWindow respectively, rather than work straight on the HWND of a panel within OWL. Besides you could use such a widget's Draw() method to trigger/refresh your own drawing. Dirk ... View more

The scheme of cross version compatibility is called "scripting", where you can run "versioned" scripts in later versions, but not the other way around. On the C++ side it takes minor efforts ( the occasional #if in single source, or version specific branches ) but here I prefer that the Adobe engineers spend their limited time on features rather than on backward emulation. Where would you cut the line? This year I've done plenty work on CS2, and support for later versions was mostly a recompile and addition of features. Anyway past year's compilers are already toast, so you will need old hardware to run them and the matching operating systems. Have you tried to purchase a copy of Metrowerks CodeWarrior for PPC recently? Plus a decent Mac capable of running that IDE? Even if you have a growing farm of oldtimers, for new developments you will prefer to take advantage of newer tools. Finally, as plugin developer you should see a certain benefits if customers are motivated to updates / upgrades every now and then, of course combined with support for all the feature improvements. Dirk ... View more

You have to maintain a private queue, without using any InDesign classes. The thread feeds the queue, and the idle task within InDesign's main thread polls it. Dirk ... View more

I'm glad you could resolve that. I would have attempted to find any more official place where the tool manager picks up the criteria to actually do the enabling, but my spare time has been very limited recently. If you have to strictly prevent text edits, also consider a TextPreProcess service. For your test cases: - select some text and open the story editor window (Cmd+Y), before you show your dialog and do your magic. If you click into that story editor window you might still be able to do text edits ... - try a double click into any text frame, rather than switching tools via tool bar. - If you have any (but text) selection, typing "T" should also choose the text tool. - Would an Undo step that undoes a text change confuse your code? - Find & Change, spell checker and so forth may also alter text. - Drag and drop of text from an outside editor, or from file system. - Drag/move, remove, duplicate etc. any XML elements in structure view. Dirk ... View more