ADC Home > Reference Library > Technical Notes > Legacy Documents > Printing >
Important: This document is part of the Legacy section of the ADC Reference Library. This information should not be used for new development.
Current information on this Reference Library topic can be found here:
|
Identifying a DTP's TypeAssociated with each LaserWriter 8 desktop printer is a LaserWriter 8.6 DTP TypesThe
|
Note: |
'=LPR'
An LPR desktop printer uses the Unix LPD protocol to communicate over TCP/IP with a print server. For more information on this protocol, please see RFC 1179 at <http://ds.internic.net/ds/dspg1intdoc.html>.
'=Cst'
The custom application desktop printer writes its PostScript to disk and then launches an application to post-process the PostScript job. See Technote 1113: Customizing the Desktop Printer Utility for additional information on custom DTPs.
'=Ird'
The PostScript job is transmitted using an infrared link to an IrDA-capable printer. LaserWriter 8 uses the IrDA specification as outlined at http://www.irda.org.
Note: |
When printing a job to a desktop printer, LaserWriter 8.6 obtains the four-byte DTP type and then looks for a shared library containing the matching hose. LaserWriter 8.6 searches for the library in the following order:
PrintingLib
file
A desktop printer's type is obtained using the SettingsLib
call psGetDTPType()
, as described in Technote 1129: The Settings Library. The Printing Plug-in Manager (as described in an upcoming Technote) is then used to find a printing plug-in of type 'hose' with a subtype matching the DTP type.
Plug-in files managed by the Printing Plug-ins Manager, such as custom hoses, are required to have a resource of type 'PLGN'
with ID -8192 that contains information about the plug-ins contained in a given file. If they do not, they cannot be used and are ignored by the LaserWriter 8.6 driver. The plug-ins are also required to have a standard 'cfrg'
resource describing the code fragments in the data fork of the file.
The 'PLGN'
resource contains information about how many shared libraries are contained in this file, and for each shared library, the type of plug-in that it is, the subtype that library handles, and the library name.
The 'PLGN'
resource is as follows:
short num; // the number of shared libraries PluginLibInfo libInfo[num]; The PluginLibInfo structure is as follows: typedef struct PluginLibInfo{ SettingsDataType type; SettingsDataSubType subtype; unsigned char libraryName[ ]; // pascal string // word aligned }PluginLibInfo; type: the type of plug-in that is described by the PluginLibInfo subtype: the subtype of data that can be handled by the plug-in described by the PluginLibInfo libraryName: the library name of the code fragment in the plug-in file described by this PluginLibInfo |
A ResEdit 'TMPL'
for editing the 'PLGN'
resource is provided in PrintingLib 8.6. The type field should be 'hose'
and the subtype should be '=XXX'
, where XXX is the custom-registered type of DTP being supported. See the Hose Type Registrations section for more details on registration. The libraryName
should match the name of the code fragment in the data fork which implements the hose.
hoseOpen
Hose fragments are required to export a single entry point. This entry point, hoseOpen()
has the following signature:
OSStatus hoseOpen(HoseInfo *hoseInfo, const BufCallbacks *callbacks, Collection hints, Handle papaH); |
The hoseOpen routine's primary job is to fill out the structure pointed to by hoseInfo.
The HoseInfo structure is filled out by a hoseOpen
procedure. The structure describes the buffer requirements of the hose as well as
the function pointers for reading, writing, and closing the hose.
typedef struct{ HoseOutProc out; // Called to write a buffer. HoseInProc in; // Called to read a buffer. HoseIdleProc idle; // Called periodically. HoseCloseProc close; // Called to shut down the connection. HoseConnProc connState; // This procedure returns the state of the // current connection. HoseStatusProc status; // Return the hose's current status string. HoseDisposeProc dispose;// The hose should free up all of its memory. Size bufSize; // The size of each allocated data buffer. long minBufs; // The hose requires this many buffers. If // there isn't enough memory to allocate them, // the client will return memFullErr. long maxBufs; // We'll never allocate more buffers than this. void *refcon; // A pointer that will be passed to the hose // routines. }HoseInfo; |
First, the hose needs to fill in the out
, in
, idle
, close
, connState
, status
, and dispose
fields with native function pointers to routines in the hose that implement the hose's functionality (i.e., the PPC code provides PPC function pointers while the 68KCFM code provides 68KCFM function pointers. No classic 68K function pointers are supported). In addition to filling in the hose's function pointers, the hose must fill in the bufSize
, minBufs
, and maxBufs
fields to describe the hose's buffer requirements. Lastly, the hose needs to fill in the refcon
field with a pointer to its own storage.
Before calling the hose to transmit data, the hose client provides buffering to improve performance. Because of this buffering, the hose must deal with only one transmit buffer and one receive buffer at a time. The hose client is responsible for allocating the buffers that it uses for providing the buffering. The hose indicates the size of the buffers its client should allocate by filling in the bufSize
field during hoseOpen
. Some typical buffer sizes include 4096 bytes for the 'PAP '
hose and 16384 bytes for the '=Fil'
hose. The hose specifies the minimum number of buffers it needs in minBufs
. It is recommended that the minBufs
field be set to at least four for bidirectional hoses and to at least two for unidirectional hoses. The maximum number of buffers to be allocated is set by the hose in the maxBufs
field. The current rule of thumb for this field is that it should not specify more than 256K worth of memory. In other words, it should be (262144 / bufSize
). This is just a general rule of thumb; some hoses may require a larger maxBufs
setting, while others will use a smaller number.
The callbacks parameter to hoseOpen
is a pointer to a BufCallbacks
structure filled in by the hose client.
typedef void (*FinishedWriteProc)(MemQElemPtr memElem, OSStatus err); typedef void (*FinishedReadProc)(MemQElemPtr memElem, OSStatus err); typedef struct{ FinishedWriteProc finishedWrite; FinishedReadProc finishedRead; }BufCallbacks; |
The hose's client places two pointers to native functions in this structure. These functions are to be called by the hose when a read or write is completed. For more information on these function pointers, see the detailed description of HoseOutProc
and HoseInProc
below. The hose should store a copy of this structure in its private data.
As part of the hoseOpen
call, the hose is handed a Collection Manager collection containing hints for configuring the job that will be transmitted through the hose. The hints in this collection are used by the QuickDraw to PostScript converter hose client to configure the byte codes that are available for its generation of PostScript language output. In particular, the kHintEighthBitTag
and kHintTransparentChannelTag
hints, as defined by LaserWriter 8.6, configure the hose client as to whether it can generate its data using byte values outside the standard PostScript language printable ASCII character set. These hints are defined as follows:
/* When generating PostScript for the output stream, the PostScript converter will by default use, if needed, characters in the range 0x80-0xFF inclusive. Use the 'kHintEighthBitTag' with a value of 'false' to prevent the converter from emitting bytes with the high bit set. */ #define kHintEighthBitTag 'bit8' #define kHintEighthBitId 1 #define kHintEighthBitVariableType Boolean #define kHintEighthBitDefault true /* When generating PostScript for the output stream, the PostScript converter will by default use, if needed, characters in the range 0x00-0x1F inclusive. Use the 'kHintTransparentChannelTag' with a value of 'false' to prevent the converter from emitting bytes less than 0x20. */ #define kHintTransparentChannelTag 'trns' #define kHintTransparentChannelId 1 #define kHintTransparentChannelVariableType Boolean #define kHintTransparentChannelDefault true |
Typically, the hints collection passed to hoseOpen
does not contain the collection items corresponding to these tag/id pairs. This is equivalent to the kHintEighthBitTag
and kHintTransparentChannelTag
hints both set to true, i.e. bytes 0x00-0xFF
are all available.
For hoses which communicate through a channel that has attributes more restrictive than these defaults, the hose must add the appropriate collection item(s) to the hints collection to ensure that the hose client only writes bytes in the supported range. Note that only hints which are more restrictive than the defaults need to be added. If a hose supports full 8-bit communications, it need not add these hints to the hints collection passed in.
In some cases, the hints collection passed to hoseOpen
already contains either or both of the kHintEighthBitTag
and kHintTransparentChannelTag
collection items and the collection item may be locked. If the hose client requires a more restrictive setting than that present, it must add the hint to the collection, regardless of whether the hint is already locked.
OSErr err; // e.g. this hose cannot transmit the high 8 bit kHintEighthBitVariableType eightBit = false; // unlock the hint if it is already there // this is OK if the hint is already unlocked err = SetCollectionItemInfo(hints, kHintEighthBitTag, kHintEighthBitId, collectionLockMask, 0); // if the hint isn't already there that's fine if(err == collectionItemNotFoundErr) err = noErr; if(!err){ err = AddCollectionItem(hints, kHintEighthBitTag, kHintEighthBitId, sizeof(eightBit), &eightBit); if(!err) err = SetCollectionItemInfo(hints, kHintEighthBitTag, kHintEighthBitId, collectionLockMask, collectionLockMask); } |
See Inside Macintosh: QuickDraw GX Environment and Utilities for more information on the Collection Manager.
The last parameter to hoseOpen
, papa
, is the handle to an extended 'PAPA'
structure. Please see Technote 1115: The Extended 'PAPA' Resource for a full description of this structure and Technote 1129: The Settings Library for the 'PAPA'
accessor routines, psPapaToCollection
and psCollectionToPapa
. Each type of hose has different communications parameters that specify the target output device and how the communications channel is to be configured. The 'PAPA'
handle provides those communications settings for the hose and are set by the creator of the desktop printer.
In its hoseOpen
routine, the hose should allocate any needed memory and begin opening the connection with the printer. The hose need not -- and for most connection types should not -- complete the connection in the hoseOpen
call. Because of the lengthy connect times of most communications techniques, the opening of the printer connection should take place asynchronously. The hose starts the connection process and then returns noErr
from hoseOpen
. The client will periodically call the hoseConnProc
requesting the current state of the connection. The hoseConnProc
looks like this:
typedef enum{ kConnClosed = 0, // Start in this state. kConnOpening, // This is the state while we wait for the // printer to accept the connection. kConnOpen, // This is the state while we do reads and // writes to the printer. kConnClosing // This is the state while we wait for the // connection to close. }ConnState; typedef ConnState (*HoseConnProc)(void *refcon); |
The hose's connection procedure should return the constant kConnOpening
while the connection is being established. Once the open completes successfully, calls to the hose's connection procedure should return kConnOpen
. If instead an error occurs while opening the hose, the hose's connection procedure should return kConnClosed
. At that point, the hose client will call the hose's HoseCloseProc
and HoseCloseProc
needs to return the appropriate error code indicating why the hose couldn't be opened.
HoseOutPro
The primary purpose of a hose is to transmit data. This is accomplished by the hose's client through calls to HoseOutProc
. HoseOutProc
looks like this:
typedef OSStatus (*HoseOutProc)(void *refcon, MemQElem *memElem); |
and thus the hose's function is:
OSStatus hoseOut(void *refcon, MemQElem *memElem); |
The refcon
parameter passed to HoseOutProc
is taken from the refcon field of the HoseInfo
structure filled out by the hoseOpen
routine. This value should be a pointer or a handle to the hose's private data.
The second parameter to HoseOutProc
, memElem
, is a pointer to a MemQElem
structure describing the data to be written.
typedef struct MemQElem{ QElemPtr qLink; // Used by Enqueue and Dequeue- private. short qType; // Our constant (kMemQueueType) to identify our // queues- private. struct BufIO *bufIO; // So we can recover buffer information- // private. Byte *buf; // Pointer to the allocated buffer. SInt32 maxBytes; // The size of the block pointed to by 'buf' SInt32 nBytes; // Number of valid bytes in 'buf'. Boolean eoj; // true if the data is followed by an end of // job. Boolean inQOnly; // This buffer should be used only for the input // routines- private. }MemQElem, *MemQElemPtr; |
Note: |
The hose client uses the 'eoj'
field of the MemQElem
structure to signal to the hose when it needs to transmit a PostScript end of job to the printer. If the 'eoj'
field of the MemQElem
structure is true, a PostScript end of job indication must be sent after or along with the buffer of data (if any) in this memElem
. For some communications channels, this 'eoj'
is a data byte sent after the data, such as control-D for serial connections. For other communications channels, such as PAP
, the end of job indicator is out of band with the data itself.
Again, it is highly recommended that hoses perform their writes and reads in an asynchronous manner. In this case, HoseOutProc
begins to write the nBytes
pointed to by buf and then returns to the caller. When the write completes, the hose signals the caller by calling the finishedWrite function pointer passed in the BufCallbacks
structure to hoseOpen
.
typedef void (*FinishedWriteProc)(MemQElem *memElem, OSStatus err); |
When making the FinishedWriteProc
call, the hose passes in the MemQElem
pointer passed to hoseOut
along with an error code. If the write finished successfully, the error code should be noErr
. If there was an error in the write, pass that code to FinishedWriteProc
.
The call to FinishedWriteProc
is an indication from the hose to the client that the hose is done with the MemQElem
structure and is ready for another hoseOut
call. In fact, the routine called through the FinishedWriteProc
may immediately make another hoseOut
call before returning to the hose. Because of this, the hose must be prepared for the hoseOut
routine to be invoked while still in an asynchronous completion routine. Furthermore, once the MemQElem
pointer is passed to FinishedWriteProc
, the hose should no longer reference it. Any data that might have been copied from the structure before calling FinishedWriteProc
is no longer valid (particularly 'buf'
). So, not only is the structure itself no longer valid, any data contained in the structure that was previously in use is also no longer valid.
HoseInProc
If a hose is managing a unidirectional communications channel, the hose need not have a routine for reading data. In this case, the hoseOpen
routine should fill in the 'in'
field of the HoseInfo
structure with NULL
during hoseOpen
.
If the hose can read data from the printer, it fills the in field of the HoseInfo
structure with a pointer to its routine that reads data.
typedef OSStatus (*HoseInProc)(void *refcon, MemQElem memElem); |
As with HoseOutProc, HoseInProc
should execute asynchronously. When invoked, HoseInProc
should start a read. When data is available and the hose's asynchronous completion routine is invoked, the hose invokes the client's FinishedReadProc
, passing back the MemQElem
pointer passed to HoseInProc
and an error code. The hose cannot read more than memElem->maxBytes bytes. In addition, the hose must fill in memElem->nBytes with the number of bytes read into memElem->buf.
typedef void (*FinishedReadProc)(MemQElem memElem, OSStatus err); |
As with data writes, once the FinishedReadProc
is called, the hose must be prepared for another call to HoseInProc
before FinishedReadProc
returns. Similar to FinishedWriteProc
, the data in the MemQElem
should be considered invalid after the FinishedReadProc
is called by the hose.
HoseIdleProc
Not all hoses are able to use asynchronous completion routines to note the end of a read or write. To help these hoses, a HoseIdleProc
can be specified in the HoseInfo
structure returned from hoseOpen
. If the idle field of the HoseInfo
structure is not NULL
, the hose's client will periodically call the HoseIdleProc
. This idle procedure can check the status of pending reads and writes and call FinishedWriteProc
and FinishedReadProc
as needed. Most hoses do not need a HoseIdleProc
, but, if one is needed, it has the following signature:
typedef OSStatus (*HoseIdleProc)(void *refcon); |
HoseStatusProc
While a hose is open, the hose's client may periodically request that the hose query the printer for status. When the call is made the hose should copy a Pascal string describing the printer's last known status into the buffer pointed to by statusStr
. The hose should also start an asyncronous status request to the printer. When the asyncronous status request returns the hose must hold that status until the next call to its status procedure.
typedef OSStatus (*HoseStatusProc)(void *refcon, StringPtr statusStr); |
Note: |
For some communications channels, such as serial channels, the status from a printer is returned on the back channel read by the HoseInProc
. In such a case, the hose's client pulls the status out of the back channel and the hose does not need to do anything other than transmit a status request to the output device.
HoseCloseProc
When the client is done with the hose, it calls the hose's HoseCloseProc
to shut down the connection. The hose should terminate any pending reads and writes and begin to shut down the connection. If this shut down procedure is immediate, it can be completed before this routine returns. If the shut down procedure takes an extended amount of time, this routine can begin the process and return.
typedef OSStatus (*HoseCloseProc)(void *refcon); |
HoseDisposeProc
After HoseCloseProc
is invoked, the HoseConnProc
is called repeatedly until kConnClosed
is returned. When the hose signals that the connection has been shut down, HoseDisposeProc
is called to allow the hose to release any memory it still holds.
typedef OSStatus (*HoseDisposeProc)(void *refcon); |
If there was an error during an asynchronous close, the HoseDisposeProc
should return a non-zero error value.
In order to ensure that we do not have conflicting hose types, we ask that you register your custom hose 4-byte type by sending an email to devprograms@apple.com. Please send the following information to register your custom hose type:
As outlined in this document, creating a custom hose for LaserWriter 8.6 is fairly straightforward and clean. Make sure that you are familiar with the other documents mentioned in this Technote before you begin on your journey to create a custom hose!
Technote 1113: Customizing the Desktop Printer Utility
Technote 1115: The Extended 'PAPA' Resource
Technote 1129: The Settings Library
Technote 1170: Printing Plug-ins Manager
Inside Macintosh: QuickDraw GX Environment and Utilities
Acrobat version of this Note (76K). |
|