mingw/html/pod/perliol.html

   1 <?xml version="1.0" ?>
   2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml">
   4 <head>
   5 <title>perliol - C API for Perl's implementation of IO in Layers.</title>
   6 <meta http-equiv="content-type" content="text/html; charset=utf-8" />
   7 <link rev="made" href="mailto:" />
   8 </head>
   9
  10 <body style="background-color: white">
  11 <table border="0" width="100%" cellspacing="0" cellpadding="3">
  12 <tr><td class="block" style="background-color: #cccccc" valign="middle">
  13 <big><strong><span class="block">&nbsp;perliol - C API for Perl's implementation of IO in Layers.</span></strong></big>
  14 </td></tr>
  15 </table>
  16
  17 <p><a name="__index__"></a></p>
  18 <!-- INDEX BEGIN -->
  19
  20 <ul>
  21
  22         <li><a href="#name">NAME</a></li>
  23         <li><a href="#synopsis">SYNOPSIS</a></li>
  24         <li><a href="#description">DESCRIPTION</a></li>
  25         <ul>
  26
  27                 <li><a href="#history_and_background">History and Background</a></li>
  28                 <li><a href="#basic_structure">Basic Structure</a></li>
  29                 <li><a href="#layers_vs_disciplines">Layers vs Disciplines</a></li>
  30                 <li><a href="#data_structures">Data Structures</a></li>
  31                 <li><a href="#functions_and_attributes">Functions and Attributes</a></li>
  32                 <li><a href="#perinstance_data">Per-instance Data</a></li>
  33                 <li><a href="#layers_in_action_">Layers in action.</a></li>
  34                 <li><a href="#perinstance_flag_bits">Per-instance flag bits</a></li>
  35                 <li><a href="#methods_in_detail">Methods in Detail</a></li>
  36                 <li><a href="#utilities">Utilities</a></li>
  37                 <li><a href="#implementing_perlio_layers">Implementing PerlIO Layers</a></li>
  38                 <li><a href="#core_layers">Core Layers</a></li>
  39                 <li><a href="#extension_layers">Extension Layers</a></li>
  40         </ul>
  41
  42         <li><a href="#todo">TODO</a></li>
  43 </ul>
  44 <!-- INDEX END -->
  45
  46 <hr />
  47 <p>
  48 </p>
  49 <h1><a name="name">NAME</a></h1>
  50 <p>perliol - C API for Perl's implementation of IO in Layers.</p>
  51 <p>
  52 </p>
  53 <hr />
  54 <h1><a name="synopsis">SYNOPSIS</a></h1>
  55 <pre>
  56     /* Defining a layer ... */
  57     #include &lt;<a href="//C|\msysgit\mingw\html/pod/perliol.html">perliol</a>.h&gt;</pre>
  58 <p>
  59 </p>
  60 <hr />
  61 <h1><a name="description">DESCRIPTION</a></h1>
  62 <p>This document describes the behavior and implementation of the PerlIO
  63 abstraction described in <a href="file://C|\msysgit\mingw\html/pod/perlapio.html">the perlapio manpage</a> when <code>USE_PERLIO</code> is defined (and
  64 <code>USE_SFIO</code> is not).</p>
  65 <p>
  66 </p>
  67 <h2><a name="history_and_background">History and Background</a></h2>
  68 <p>The PerlIO abstraction was introduced in perl5.003_02 but languished as
  69 just an abstraction until perl5.7.0. However during that time a number
  70 of perl extensions switched to using it, so the API is mostly fixed to
  71 maintain (source) compatibility.</p>
  72 <p>The aim of the implementation is to provide the PerlIO API in a flexible
  73 and platform neutral manner. It is also a trial of an ``Object Oriented
  74 C, with vtables'' approach which may be applied to perl6.</p>
  75 <p>
  76 </p>
  77 <h2><a name="basic_structure">Basic Structure</a></h2>
  78 <p>PerlIO is a stack of layers.</p>
  79 <p>The low levels of the stack work with the low-level operating system
  80 calls (file descriptors in C) getting bytes in and out, the higher
  81 layers of the stack buffer, filter, and otherwise manipulate the I/O,
  82 and return characters (or bytes) to Perl.  Terms <em>above</em> and <em>below</em>
  83 are used to refer to the relative positioning of the stack layers.</p>
  84 <p>A layer contains a ``vtable'', the table of I/O operations (at C level
  85 a table of function pointers), and status flags.  The functions in the
  86 vtable implement operations like ``open'', ``read'', and ``write''.</p>
  87 <p>When I/O, for example ``read'', is requested, the request goes from Perl
  88 first down the stack using ``read'' functions of each layer, then at the
  89 bottom the input is requested from the operating system services, then
  90 the result is returned up the stack, finally being interpreted as Perl
  91 data.</p>
  92 <p>The requests do not necessarily go always all the way down to the
  93 operating system: that's where PerlIO buffering comes into play.</p>
  94 <p>When you do an <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open()</code></a> and specify extra PerlIO layers to be deployed,
  95 the layers you specify are ``pushed'' on top of the already existing
  96 default stack.  One way to see it is that ``operating system is
  97 on the left'' and ``Perl is on the right''.</p>
  98 <p>What exact layers are in this default stack depends on a lot of
  99 things: your operating system, Perl version, Perl compile time
 100 configuration, and Perl runtime configuration.  See <a href="file://C|\msysgit\mingw\html/lib/Encode/PerlIO.html">the PerlIO manpage</a>,
 101 <a href="file://C|\msysgit\mingw\html/pod/perlrun.html#perlio">PERLIO in the perlrun manpage</a>, and <a href="file://C|\msysgit\mingw\html/lib/open.html">the open manpage</a> for more information.</p>
 102 <p><a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_binmode"><code>binmode()</code></a> operates similarly to open(): by default the specified
 103 layers are pushed on top of the existing stack.</p>
 104 <p>However, note that even as the specified layers are ``pushed on top''
 105 for <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open()</code></a> and binmode(), this doesn't mean that the effects are
 106 limited to the ``top'': PerlIO layers can be very 'active' and inspect
 107 and affect layers also deeper in the stack.  As an example there
 108 is a layer called ``raw'' which repeatedly ``pops'' layers until
 109 it reaches the first layer that has declared itself capable of
 110 handling binary data.  The ``pushed'' layers are processed in left-to-right
 111 order.</p>
 112 <p><a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_sysopen"><code>sysopen()</code></a> operates (unsurprisingly) at a lower level in the stack than
 113 open().  For example in UNIX or UNIX-like systems <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_sysopen"><code>sysopen()</code></a> operates
 114 directly at the level of file descriptors: in the terms of PerlIO
 115 layers, it uses only the ``unix'' layer, which is a rather thin wrapper
 116 on top of the UNIX file descriptors.</p>
 117 <p>
 118 </p>
 119 <h2><a name="layers_vs_disciplines">Layers vs Disciplines</a></h2>
 120 <p>Initial discussion of the ability to modify IO streams behaviour used
 121 the term ``discipline'' for the entities which were added. This came (I
 122 believe) from the use of the term in ``sfio'', which in turn borrowed it
 123 from ``line disciplines'' on Unix terminals. However, this document (and
 124 the C code) uses the term ``layer''.</p>
 125 <p>This is, I hope, a natural term given the implementation, and should
 126 avoid connotations that are inherent in earlier uses of ``discipline''
 127 for things which are rather different.</p>
 128 <p>
 129 </p>
 130 <h2><a name="data_structures">Data Structures</a></h2>
 131 <p>The basic data structure is a PerlIOl:</p>
 132 <pre>
 133         typedef struct _PerlIO PerlIOl;
 134         typedef struct _PerlIO_funcs PerlIO_funcs;
 135         typedef PerlIOl *PerlIO;</pre>
 136 <pre>
 137         struct _PerlIO
 138         {
 139          PerlIOl *      next;       /* Lower layer */
 140          PerlIO_funcs * tab;        /* Functions for this layer */
 141          IV             flags;      /* Various flags for state */
 142         };</pre>
 143 <p>A <code>PerlIOl *</code> is a pointer to the struct, and the <em>application</em>
 144 level <code>PerlIO *</code> is a pointer to a <code>PerlIOl *</code> - i.e. a pointer
 145 to a pointer to the struct. This allows the application level <code>PerlIO *</code>
 146 to remain constant while the actual <code>PerlIOl *</code> underneath
 147 changes. (Compare perl's <code>SV *</code> which remains constant while its
 148 <code>sv_any</code> field changes as the scalar's type changes.) An IO stream is
 149 then in general represented as a pointer to this linked-list of
 150 ``layers''.</p>
 151 <p>It should be noted that because of the double indirection in a <code>PerlIO *</code>,
 152 a <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_next"><code>&amp;(perlio-&gt;next)</code></a> ``is'' a <code>PerlIO *</code>, and so to some degree
 153 at least one layer can use the ``standard'' API on the next layer down.</p>
 154 <p>A ``layer'' is composed of two parts:</p>
 155 <ol>
 156 <li>
 157 <p>The functions and attributes of the ``layer class''.</p>
 158 </li>
 159 <li>
 160 <p>The per-instance data for a particular handle.</p>
 161 </li>
 162 </ol>
 163 <p>
 164 </p>
 165 <h2><a name="functions_and_attributes">Functions and Attributes</a></h2>
 166 <p>The functions and attributes are accessed via the ``tab'' (for table)
 167 member of <code>PerlIOl</code>. The functions (methods of the layer ``class'') are
 168 fixed, and are defined by the <code>PerlIO_funcs</code> type. They are broadly the
 169 same as the public <code>PerlIO_xxxxx</code> functions:</p>
 170 <pre>
 171   struct _PerlIO_funcs
 172   {
 173    Size_t               fsize;
 174    char *               name;
 175    Size_t               size;
 176    IV           kind;
 177    IV           (*Pushed)(pTHX_ PerlIO *f,const char *mode,SV *arg, PerlIO_funcs *tab);
 178    IV           (*Popped)(pTHX_ PerlIO *f);
 179    PerlIO *     (*Open)(pTHX_ PerlIO_funcs *tab,
 180                         AV *layers, IV n,
 181                         const char *mode,
 182                         int fd, int imode, int perm,
 183                         PerlIO *old,
 184                         int narg, SV **args);
 185    IV           (*Binmode)(pTHX_ PerlIO *f);
 186    SV *         (*Getarg)(pTHX_ PerlIO *f, CLONE_PARAMS *param, int flags)
 187    IV           (*Fileno)(pTHX_ PerlIO *f);
 188    PerlIO *     (*Dup)(pTHX_ PerlIO *f, PerlIO *o, CLONE_PARAMS *param, int flags)
 189    /* Unix-like functions - cf sfio line disciplines */
 190    SSize_t      (*Read)(pTHX_ PerlIO *f, void *vbuf, Size_t count);
 191    SSize_t      (*Unread)(pTHX_ PerlIO *f, const void *vbuf, Size_t count);
 192    SSize_t      (*Write)(pTHX_ PerlIO *f, const void *vbuf, Size_t count);
 193    IV           (*Seek)(pTHX_ PerlIO *f, Off_t offset, int whence);
 194    Off_t        (*Tell)(pTHX_ PerlIO *f);
 195    IV           (*Close)(pTHX_ PerlIO *f);
 196    /* Stdio-like buffered IO functions */
 197    IV           (*Flush)(pTHX_ PerlIO *f);
 198    IV           (*Fill)(pTHX_ PerlIO *f);
 199    IV           (*Eof)(pTHX_ PerlIO *f);
 200    IV           (*Error)(pTHX_ PerlIO *f);
 201    void         (*Clearerr)(pTHX_ PerlIO *f);
 202    void         (*Setlinebuf)(pTHX_ PerlIO *f);
 203    /* Perl's snooping functions */
 204    STDCHAR *    (*Get_base)(pTHX_ PerlIO *f);
 205    Size_t       (*Get_bufsiz)(pTHX_ PerlIO *f);
 206    STDCHAR *    (*Get_ptr)(pTHX_ PerlIO *f);
 207    SSize_t      (*Get_cnt)(pTHX_ PerlIO *f);
 208    void         (*Set_ptrcnt)(pTHX_ PerlIO *f,STDCHAR *ptr,SSize_t cnt);
 209   };</pre>
 210 <p>The first few members of the struct give a function table size for
 211 compatibility check ``name'' for the layer, the  size to <code>malloc</code> for the per-instance data,
 212 and some flags which are attributes of the class as whole (such as whether it is a buffering
 213 layer), then follow the functions which fall into four basic groups:</p>
 214 <ol>
 215 <li>
 216 <p>Opening and setup functions</p>
 217 </li>
 218 <li>
 219 <p>Basic IO operations</p>
 220 </li>
 221 <li>
 222 <p>Stdio class buffering options.</p>
 223 </li>
 224 <li>
 225 <p>Functions to support Perl's traditional ``fast'' access to the buffer.</p>
 226 </li>
 227 </ol>
 228 <p>A layer does not have to implement all the functions, but the whole
 229 table has to be present. Unimplemented slots can be NULL (which will
 230 result in an error when called) or can be filled in with stubs to
 231 ``inherit'' behaviour from a ``base class''. This ``inheritance'' is fixed
 232 for all instances of the layer, but as the layer chooses which stubs
 233 to populate the table, limited ``multiple inheritance'' is possible.</p>
 234 <p>
 235 </p>
 236 <h2><a name="perinstance_data">Per-instance Data</a></h2>
 237 <p>The per-instance data are held in memory beyond the basic PerlIOl
 238 struct, by making a PerlIOl the first member of the layer's struct
 239 thus:</p>
 240 <pre>
 241         typedef struct
 242         {
 243          struct _PerlIO base;       /* Base &quot;class&quot; info */
 244          STDCHAR *      buf;        /* Start of buffer */
 245          STDCHAR *      end;        /* End of valid part of buffer */
 246          STDCHAR *      ptr;        /* Current position in buffer */
 247          Off_t          posn;       /* Offset of buf into the file */
 248          Size_t         bufsiz;     /* Real size of buffer */
 249          IV             oneword;    /* Emergency buffer */
 250         } PerlIOBuf;</pre>
 251 <p>In this way (as for perl's scalars) a pointer to a PerlIOBuf can be
 252 treated as a pointer to a PerlIOl.</p>
 253 <p>
 254 </p>
 255 <h2><a name="layers_in_action_">Layers in action.</a></h2>
 256 <pre>
 257                 table           perlio          unix
 258             |           |
 259             +-----------+    +----------+    +--------+
 260    PerlIO -&gt;|           |---&gt;|  next    |---&gt;|  NULL  |
 261             +-----------+    +----------+    +--------+
 262             |           |    |  buffer  |    |   fd   |
 263             +-----------+    |          |    +--------+
 264             |           |    +----------+</pre>
 265 <p>The above attempts to show how the layer scheme works in a simple case.
 266 The application's <code>PerlIO *</code> points to an entry in the <code>table(s)</code>
 267 representing open (allocated) handles. For example the first three slots
 268 in the table correspond to <code>stdin</code>,<code>stdout</code> and <code>stderr</code>. The table
 269 in turn points to the current ``top'' layer for the handle - in this case
 270 an instance of the generic buffering layer ``perlio''. That layer in turn
 271 points to the next layer down - in this case the lowlevel ``unix'' layer.</p>
 272 <p>The above is roughly equivalent to a ``stdio'' buffered stream, but with
 273 much more flexibility:</p>
 274 <ul>
 275 <li>
 276 <p>If Unix level <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_read"><code>read</code></a>/<a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_write"><code>write</code></a>/<code>lseek</code> is not appropriate for (say)
 277 sockets then the ``unix'' layer can be replaced (at open time or even
 278 dynamically) with a ``socket'' layer.</p>
 279 </li>
 280 <li>
 281 <p>Different handles can have different buffering schemes. The ``top''
 282 layer could be the ``mmap'' layer if reading disk files was quicker
 283 using <code>mmap</code> than <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_read"><code>read</code></a>. An ``unbuffered'' stream can be implemented
 284 simply by not having a buffer layer.</p>
 285 </li>
 286 <li>
 287 <p>Extra layers can be inserted to process the data as it flows through.
 288 This was the driving need for including the scheme in perl 5.7.0+ - we
 289 needed a mechanism to allow data to be translated between perl's
 290 internal encoding (conceptually at least Unicode as UTF-8), and the
 291 ``native'' format used by the system. This is provided by the
 292 ``:encoding(xxxx)'' layer which typically sits above the buffering layer.</p>
 293 </li>
 294 <li>
 295 <p>A layer can be added that does ``\n'' to CRLF translation. This layer
 296 can be used on any platform, not just those that normally do such
 297 things.</p>
 298 </li>
 299 </ul>
 300 <p>
 301 </p>
 302 <h2><a name="perinstance_flag_bits">Per-instance flag bits</a></h2>
 303 <p>The generic flag bits are a hybrid of <code>O_XXXXX</code> style flags deduced
 304 from the mode string passed to <code>PerlIO_open()</code>, and state bits for
 305 typical buffer layers.</p>
 306 <dl>
 307 <dt><strong><a name="item_perlio_f_eof">PERLIO_F_EOF</a></strong>
 308
 309 <dd>
 310 <p>End of file.</p>
 311 </dd>
 312 </li>
 313 <dt><strong><a name="item_perlio_f_canwrite">PERLIO_F_CANWRITE</a></strong>
 314
 315 <dd>
 316 <p>Writes are permitted, i.e. opened as ``w'' or ``r+'' or ``a'', etc.</p>
 317 </dd>
 318 </li>
 319 <dt><strong><a name="item_perlio_f_canread">PERLIO_F_CANREAD</a></strong>
 320
 321 <dd>
 322 <p>Reads are permitted i.e. opened ``r'' or ``w+'' (or even ``a+'' - ick).</p>
 323 </dd>
 324 </li>
 325 <dt><strong><a name="item_perlio_f_error">PERLIO_F_ERROR</a></strong>
 326
 327 <dd>
 328 <p>An error has occurred (for <code>PerlIO_error()</code>).</p>
 329 </dd>
 330 </li>
 331 <dt><strong><a name="item_perlio_f_truncate">PERLIO_F_TRUNCATE</a></strong>
 332
 333 <dd>
 334 <p>Truncate file suggested by open mode.</p>
 335 </dd>
 336 </li>
 337 <dt><strong><a name="item_perlio_f_append">PERLIO_F_APPEND</a></strong>
 338
 339 <dd>
 340 <p>All writes should be appends.</p>
 341 </dd>
 342 </li>
 343 <dt><strong><a name="item_perlio_f_crlf">PERLIO_F_CRLF</a></strong>
 344
 345 <dd>
 346 <p>Layer is performing Win32-like ``\n'' mapped to CR,LF for output and CR,LF
 347 mapped to ``\n'' for input. Normally the provided ``crlf'' layer is the only
 348 layer that need bother about this. <code>PerlIO_binmode()</code> will mess with this
 349 flag rather than add/remove layers if the <a href="#item_perlio_k_cancrlf"><code>PERLIO_K_CANCRLF</code></a> bit is set
 350 for the layers class.</p>
 351 </dd>
 352 </li>
 353 <dt><strong><a name="item_perlio_f_utf8">PERLIO_F_UTF8</a></strong>
 354
 355 <dd>
 356 <p>Data written to this layer should be UTF-8 encoded; data provided
 357 by this layer should be considered UTF-8 encoded. Can be set on any layer
 358 by ``:utf8'' dummy layer. Also set on ``:encoding'' layer.</p>
 359 </dd>
 360 </li>
 361 <dt><strong><a name="item_perlio_f_unbuf">PERLIO_F_UNBUF</a></strong>
 362
 363 <dd>
 364 <p>Layer is unbuffered - i.e. write to next layer down should occur for
 365 each write to this layer.</p>
 366 </dd>
 367 </li>
 368 <dt><strong><a name="item_perlio_f_wrbuf">PERLIO_F_WRBUF</a></strong>
 369
 370 <dd>
 371 <p>The buffer for this layer currently holds data written to it but not sent
 372 to next layer.</p>
 373 </dd>
 374 </li>
 375 <dt><strong><a name="item_perlio_f_rdbuf">PERLIO_F_RDBUF</a></strong>
 376
 377 <dd>
 378 <p>The buffer for this layer currently holds unconsumed data read from
 379 layer below.</p>
 380 </dd>
 381 </li>
 382 <dt><strong><a name="item_perlio_f_linebuf">PERLIO_F_LINEBUF</a></strong>
 383
 384 <dd>
 385 <p>Layer is line buffered. Write data should be passed to next layer down
 386 whenever a ``\n'' is seen. Any data beyond the ``\n'' should then be
 387 processed.</p>
 388 </dd>
 389 </li>
 390 <dt><strong><a name="item_perlio_f_temp">PERLIO_F_TEMP</a></strong>
 391
 392 <dd>
 393 <p>File has been <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_unlink"><code>unlink()</code></a>ed, or should be deleted on <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_close"><code>close()</code></a>.</p>
 394 </dd>
 395 </li>
 396 <dt><strong><a name="item_perlio_f_open">PERLIO_F_OPEN</a></strong>
 397
 398 <dd>
 399 <p>Handle is open.</p>
 400 </dd>
 401 </li>
 402 <dt><strong><a name="item_perlio_f_fastgets">PERLIO_F_FASTGETS</a></strong>
 403
 404 <dd>
 405 <p>This instance of this layer supports the ``fast <code>gets</code>'' interface.
 406 Normally set based on <a href="#item_perlio_k_fastgets"><code>PERLIO_K_FASTGETS</code></a> for the class and by the
 407 existence of the <code>function(s)</code> in the table. However a class that
 408 normally provides that interface may need to avoid it on a
 409 particular instance. The ``pending'' layer needs to do this when
 410 it is pushed above a layer which does not support the interface.
 411 (Perl's <code>sv_gets()</code> does not expect the streams fast <code>gets</code> behaviour
 412 to change during one ``get''.)</p>
 413 </dd>
 414 </li>
 415 </dl>
 416 <p>
 417 </p>
 418 <h2><a name="methods_in_detail">Methods in Detail</a></h2>
 419 <dl>
 420 <dt><strong><a name="item_fsize">fsize</a></strong>
 421
 422 <dd>
 423 <pre>
 424         Size_t fsize;</pre>
 425 </dd>
 426 <dd>
 427 <p>Size of the function table. This is compared against the value PerlIO
 428 code ``knows'' as a compatibility check. Future versions <em>may</em> be able
 429 to tolerate layers compiled against an old version of the headers.</p>
 430 </dd>
 431 <dt><strong><a name="item_name">name</a></strong>
 432
 433 <dd>
 434 <pre>
 435         char * name;</pre>
 436 </dd>
 437 <dd>
 438 <p>The name of the layer whose <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open()</code></a> method Perl should invoke on
 439 open().  For example if the layer is called APR, you will call:</p>
 440 </dd>
 441 <dd>
 442 <pre>
 443   open $fh, &quot;&gt;:APR&quot;, ...</pre>
 444 </dd>
 445 <dd>
 446 <p>and Perl knows that it has to invoke the <code>PerlIOAPR_open()</code> method
 447 implemented by the APR layer.</p>
 448 </dd>
 449 <dt><strong><a name="item_size">size</a></strong>
 450
 451 <dd>
 452 <pre>
 453         Size_t size;</pre>
 454 </dd>
 455 <dd>
 456 <p>The size of the per-instance data structure, e.g.:</p>
 457 </dd>
 458 <dd>
 459 <pre>
 460   sizeof(PerlIOAPR)</pre>
 461 </dd>
 462 <dd>
 463 <p>If this field is zero then <code>PerlIO_pushed</code> does not malloc anything
 464 and assumes layer's Pushed function will do any required layer stack
 465 manipulation - used to avoid malloc/free overhead for dummy layers.
 466 If the field is non-zero it must be at least the size of <code>PerlIOl</code>,
 467 <code>PerlIO_pushed</code> will allocate memory for the layer's data structures
 468 and link new layer onto the stream's stack. (If the layer's Pushed
 469 method returns an error indication the layer is popped again.)</p>
 470 </dd>
 471 <dt><strong><a name="item_kind">kind</a></strong>
 472
 473 <dd>
 474 <pre>
 475         IV kind;</pre>
 476 </dd>
 477 <ul>
 478 <li><strong><a name="item_perlio_k_buffered">PERLIO_K_BUFFERED</a></strong>
 479
 480 <p>The layer is buffered.</p>
 481 </li>
 482 <li><strong><a name="item_perlio_k_raw">PERLIO_K_RAW</a></strong>
 483
 484 <p>The layer is acceptable to have in a <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_binmode"><code>binmode(FH)</code></a> stack - i.e. it does not
 485 (or will configure itself not to) transform bytes passing through it.</p>
 486 </li>
 487 <li><strong><a name="item_perlio_k_cancrlf">PERLIO_K_CANCRLF</a></strong>
 488
 489 <p>Layer can translate between ``\n'' and CRLF line ends.</p>
 490 </li>
 491 <li><strong><a name="item_perlio_k_fastgets">PERLIO_K_FASTGETS</a></strong>
 492
 493 <p>Layer allows buffer snooping.</p>
 494 </li>
 495 <li><strong><a name="item_perlio_k_multiarg">PERLIO_K_MULTIARG</a></strong>
 496
 497 <p>Used when the layer's <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open()</code></a> accepts more arguments than usual. The
 498 extra arguments should come not before the <code>MODE</code> argument. When this
 499 flag is used it's up to the layer to validate the args.</p>
 500 </li>
 501 </ul>
 502 <dt><strong><a name="item_pushed">Pushed</a></strong>
 503
 504 <dd>
 505 <pre>
 506         IV      (*Pushed)(pTHX_ PerlIO *f,const char *mode, SV *arg);</pre>
 507 </dd>
 508 <dd>
 509 <p>The only absolutely mandatory method. Called when the layer is pushed
 510 onto the stack.  The <code>mode</code> argument may be NULL if this occurs
 511 post-open. The <code>arg</code> will be non-<code>NULL</code> if an argument string was
 512 passed. In most cases this should call <code>PerlIOBase_pushed()</code> to
 513 convert <code>mode</code> into the appropriate <code>PERLIO_F_XXXXX</code> flags in
 514 addition to any actions the layer itself takes.  If a layer is not
 515 expecting an argument it need neither save the one passed to it, nor
 516 provide <a href="#item_getarg"><code>Getarg()</code></a> (it could perhaps <code>Perl_warn</code> that the argument
 517 was un-expected).</p>
 518 </dd>
 519 <dd>
 520 <p>Returns 0 on success. On failure returns -1 and should set errno.</p>
 521 </dd>
 522 <dt><strong><a name="item_popped">Popped</a></strong>
 523
 524 <dd>
 525 <pre>
 526         IV      (*Popped)(pTHX_ PerlIO *f);</pre>
 527 </dd>
 528 <dd>
 529 <p>Called when the layer is popped from the stack. A layer will normally
 530 be popped after <a href="#item_close"><code>Close()</code></a> is called. But a layer can be popped
 531 without being closed if the program is dynamically managing layers on
 532 the stream. In such cases <a href="#item_popped"><code>Popped()</code></a> should free any resources
 533 (buffers, translation tables, ...) not held directly in the layer's
 534 struct.  It should also <a href="#item_unread"><code>Unread()</code></a> any unconsumed data that has been
 535 read and buffered from the layer below back to that layer, so that it
 536 can be re-provided to what ever is now above.</p>
 537 </dd>
 538 <dd>
 539 <p>Returns 0 on success and failure.  If <a href="#item_popped"><code>Popped()</code></a> returns <em>true</em> then
 540 <em>perlio.c</em> assumes that either the layer has popped itself, or the
 541 layer is super special and needs to be retained for other reasons.
 542 In most cases it should return <em>false</em>.</p>
 543 </dd>
 544 <dt><strong><a name="item_open">Open</a></strong>
 545
 546 <dd>
 547 <pre>
 548         PerlIO *        (*Open)(...);</pre>
 549 </dd>
 550 <dd>
 551 <p>The <a href="#item_open"><code>Open()</code></a> method has lots of arguments because it combines the
 552 functions of perl's <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a>, <code>PerlIO_open</code>, perl's <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_sysopen"><code>sysopen</code></a>,
 553 <code>PerlIO_fdopen</code> and <code>PerlIO_reopen</code>.  The full prototype is as
 554 follows:</p>
 555 </dd>
 556 <dd>
 557 <pre>
 558  PerlIO *       (*Open)(pTHX_ PerlIO_funcs *tab,
 559                         AV *layers, IV n,
 560                         const char *mode,
 561                         int fd, int imode, int perm,
 562                         PerlIO *old,
 563                         int narg, SV **args);</pre>
 564 </dd>
 565 <dd>
 566 <p>Open should (perhaps indirectly) call <code>PerlIO_allocate()</code> to allocate
 567 a slot in the table and associate it with the layers information for
 568 the opened file, by calling <code>PerlIO_push</code>.  The <em>layers</em> AV is an
 569 array of all the layers destined for the <code>PerlIO *</code>, and any
 570 arguments passed to them, <em>n</em> is the index into that array of the
 571 layer being called. The macro <code>PerlIOArg</code> will return a (possibly
 572 <code>NULL</code>) SV * for the argument passed to the layer.</p>
 573 </dd>
 574 <dd>
 575 <p>The <em>mode</em> string is an ``<code>fopen()</code>-like'' string which would match
 576 the regular expression <code>/^[I#]?[rwa]\+?[bt]?$/</code>.</p>
 577 </dd>
 578 <dd>
 579 <p>The <code>'I'</code> prefix is used during creation of <code>stdin</code>..<code>stderr</code> via
 580 special <code>PerlIO_fdopen</code> calls; the <code>'#'</code> prefix means that this is
 581 <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_sysopen"><code>sysopen</code></a> and that <em>imode</em> and <em>perm</em> should be passed to
 582 <code>PerlLIO_open3</code>; <code>'r'</code> means <strong>r</strong>ead, <code>'w'</code> means <strong>w</strong>rite and
 583 <code>'a'</code> means <strong>a</strong>ppend. The <code>'+'</code> suffix means that both reading and
 584 writing/appending are permitted.  The <code>'b'</code> suffix means file should
 585 be binary, and <code>'t'</code> means it is text. (Almost all layers should do
 586 the IO in binary mode, and ignore the b/t bits. The <code>:crlf</code> layer
 587 should be pushed to handle the distinction.)</p>
 588 </dd>
 589 <dd>
 590 <p>If <em>old</em> is not <code>NULL</code> then this is a <code>PerlIO_reopen</code>. Perl itself
 591 does not use this (yet?) and semantics are a little vague.</p>
 592 </dd>
 593 <dd>
 594 <p>If <em>fd</em> not negative then it is the numeric file descriptor <em>fd</em>,
 595 which will be open in a manner compatible with the supplied mode
 596 string, the call is thus equivalent to <code>PerlIO_fdopen</code>. In this case
 597 <em>nargs</em> will be zero.</p>
 598 </dd>
 599 <dd>
 600 <p>If <em>nargs</em> is greater than zero then it gives the number of arguments
 601 passed to <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a>, otherwise it will be 1 if for example
 602 <code>PerlIO_open</code> was called.  In simple cases <code>SvPV_nolen(*args)</code> is the
 603 pathname to open.</p>
 604 </dd>
 605 <dd>
 606 <p>Having said all that translation-only layers do not need to provide
 607 <a href="#item_open"><code>Open()</code></a> at all, but rather leave the opening to a lower level layer
 608 and wait to be ``pushed''.  If a layer does provide <a href="#item_open"><code>Open()</code></a> it should
 609 normally call the <a href="#item_open"><code>Open()</code></a> method of next layer down (if any) and
 610 then push itself on top if that succeeds.</p>
 611 </dd>
 612 <dd>
 613 <p>If <code>PerlIO_push</code> was performed and open has failed, it must
 614 <code>PerlIO_pop</code> itself, since if it's not, the layer won't be removed
 615 and may cause bad problems.</p>
 616 </dd>
 617 <dd>
 618 <p>Returns <code>NULL</code> on failure.</p>
 619 </dd>
 620 <dt><strong><a name="item_binmode">Binmode</a></strong>
 621
 622 <dd>
 623 <pre>
 624         IV        (*Binmode)(pTHX_ PerlIO *f);</pre>
 625 </dd>
 626 <dd>
 627 <p>Optional. Used when <code>:raw</code> layer is pushed (explicitly or as a result
 628 of binmode(FH)). If not present layer will be popped. If present
 629 should configure layer as binary (or pop itself) and return 0.
 630 If it returns -1 for error <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_binmode"><code>binmode</code></a> will fail with layer
 631 still on the stack.</p>
 632 </dd>
 633 <dt><strong><a name="item_getarg">Getarg</a></strong>
 634
 635 <dd>
 636 <pre>
 637         SV *      (*Getarg)(pTHX_ PerlIO *f,
 638                             CLONE_PARAMS *param, int flags);</pre>
 639 </dd>
 640 <dd>
 641 <p>Optional. If present should return an SV * representing the string
 642 argument passed to the layer when it was
 643 pushed. e.g. ``:encoding(ascii)'' would return an SvPV with value
 644 ``ascii''. (<em>param</em> and <em>flags</em> arguments can be ignored in most
 645 cases)</p>
 646 </dd>
 647 <dd>
 648 <p><a href="#item_dup"><code>Dup</code></a> uses <a href="#item_getarg"><code>Getarg</code></a> to retrieve the argument originally passed to
 649 <a href="#item_pushed"><code>Pushed</code></a>, so you must implement this function if your layer has an
 650 extra argument to <a href="#item_pushed"><code>Pushed</code></a> and will ever be <a href="#item_dup"><code>Dup</code></a>ed.</p>
 651 </dd>
 652 <dt><strong><a name="item_fileno">Fileno</a></strong>
 653
 654 <dd>
 655 <pre>
 656         IV        (*Fileno)(pTHX_ PerlIO *f);</pre>
 657 </dd>
 658 <dd>
 659 <p>Returns the Unix/Posix numeric file descriptor for the handle. Normally
 660 <code>PerlIOBase_fileno()</code> (which just asks next layer down) will suffice
 661 for this.</p>
 662 </dd>
 663 <dd>
 664 <p>Returns -1 on error, which is considered to include the case where the
 665 layer cannot provide such a file descriptor.</p>
 666 </dd>
 667 <dt><strong><a name="item_dup">Dup</a></strong>
 668
 669 <dd>
 670 <pre>
 671         PerlIO * (*Dup)(pTHX_ PerlIO *f, PerlIO *o,
 672                         CLONE_PARAMS *param, int flags);</pre>
 673 </dd>
 674 <dd>
 675 <p>XXX: Needs more docs.</p>
 676 </dd>
 677 <dd>
 678 <p>Used as part of the ``clone'' process when a thread is spawned (in which
 679 case param will be non-NULL) and when a stream is being duplicated via
 680 '&amp;' in the <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a>.</p>
 681 </dd>
 682 <dd>
 683 <p>Similar to <a href="#item_open"><code>Open</code></a>, returns PerlIO* on success, <code>NULL</code> on failure.</p>
 684 </dd>
 685 <dt><strong><a name="item_read">Read</a></strong>
 686
 687 <dd>
 688 <pre>
 689         SSize_t (*Read)(pTHX_ PerlIO *f, void *vbuf, Size_t count);</pre>
 690 </dd>
 691 <dd>
 692 <p>Basic read operation.</p>
 693 </dd>
 694 <dd>
 695 <p>Typically will call <a href="#item_fill"><code>Fill</code></a> and manipulate pointers (possibly via the
 696 API).  <code>PerlIOBuf_read()</code> may be suitable for derived classes which
 697 provide ``fast gets'' methods.</p>
 698 </dd>
 699 <dd>
 700 <p>Returns actual bytes read, or -1 on an error.</p>
 701 </dd>
 702 <dt><strong><a name="item_unread">Unread</a></strong>
 703
 704 <dd>
 705 <pre>
 706         SSize_t (*Unread)(pTHX_ PerlIO *f,
 707                           const void *vbuf, Size_t count);</pre>
 708 </dd>
 709 <dd>
 710 <p>A superset of stdio's <code>ungetc()</code>. Should arrange for future reads to
 711 see the bytes in <code>vbuf</code>. If there is no obviously better implementation
 712 then <code>PerlIOBase_unread()</code> provides the function by pushing a ``fake''
 713 ``pending'' layer above the calling layer.</p>
 714 </dd>
 715 <dd>
 716 <p>Returns the number of unread chars.</p>
 717 </dd>
 718 <dt><strong><a name="item_write">Write</a></strong>
 719
 720 <dd>
 721 <pre>
 722         SSize_t (*Write)(PerlIO *f, const void *vbuf, Size_t count);</pre>
 723 </dd>
 724 <dd>
 725 <p>Basic write operation.</p>
 726 </dd>
 727 <dd>
 728 <p>Returns bytes written or -1 on an error.</p>
 729 </dd>
 730 <dt><strong><a name="item_seek">Seek</a></strong>
 731
 732 <dd>
 733 <pre>
 734         IV      (*Seek)(pTHX_ PerlIO *f, Off_t offset, int whence);</pre>
 735 </dd>
 736 <dd>
 737 <p>Position the file pointer. Should normally call its own <a href="#item_flush"><code>Flush</code></a>
 738 method and then the <a href="#item_seek"><code>Seek</code></a> method of next layer down.</p>
 739 </dd>
 740 <dd>
 741 <p>Returns 0 on success, -1 on failure.</p>
 742 </dd>
 743 <dt><strong><a name="item_tell">Tell</a></strong>
 744
 745 <dd>
 746 <pre>
 747         Off_t   (*Tell)(pTHX_ PerlIO *f);</pre>
 748 </dd>
 749 <dd>
 750 <p>Return the file pointer. May be based on layers cached concept of
 751 position to avoid overhead.</p>
 752 </dd>
 753 <dd>
 754 <p>Returns -1 on failure to get the file pointer.</p>
 755 </dd>
 756 <dt><strong><a name="item_close">Close</a></strong>
 757
 758 <dd>
 759 <pre>
 760         IV      (*Close)(pTHX_ PerlIO *f);</pre>
 761 </dd>
 762 <dd>
 763 <p>Close the stream. Should normally call <code>PerlIOBase_close()</code> to flush
 764 itself and close layers below, and then deallocate any data structures
 765 (buffers, translation tables, ...) not  held directly in the data
 766 structure.</p>
 767 </dd>
 768 <dd>
 769 <p>Returns 0 on success, -1 on failure.</p>
 770 </dd>
 771 <dt><strong><a name="item_flush">Flush</a></strong>
 772
 773 <dd>
 774 <pre>
 775         IV      (*Flush)(pTHX_ PerlIO *f);</pre>
 776 </dd>
 777 <dd>
 778 <p>Should make stream's state consistent with layers below. That is, any
 779 buffered write data should be written, and file position of lower layers
 780 adjusted for data read from below but not actually consumed.
 781 (Should perhaps <a href="#item_unread"><code>Unread()</code></a> such data to the lower layer.)</p>
 782 </dd>
 783 <dd>
 784 <p>Returns 0 on success, -1 on failure.</p>
 785 </dd>
 786 <dt><strong><a name="item_fill">Fill</a></strong>
 787
 788 <dd>
 789 <pre>
 790         IV      (*Fill)(pTHX_ PerlIO *f);</pre>
 791 </dd>
 792 <dd>
 793 <p>The buffer for this layer should be filled (for read) from layer
 794 below.  When you ``subclass'' PerlIOBuf layer, you want to use its
 795 <em>_read</em> method and to supply your own fill method, which fills the
 796 PerlIOBuf's buffer.</p>
 797 </dd>
 798 <dd>
 799 <p>Returns 0 on success, -1 on failure.</p>
 800 </dd>
 801 <dt><strong><a name="item_eof">Eof</a></strong>
 802
 803 <dd>
 804 <pre>
 805         IV      (*Eof)(pTHX_ PerlIO *f);</pre>
 806 </dd>
 807 <dd>
 808 <p>Return end-of-file indicator. <code>PerlIOBase_eof()</code> is normally sufficient.</p>
 809 </dd>
 810 <dd>
 811 <p>Returns 0 on end-of-file, 1 if not end-of-file, -1 on error.</p>
 812 </dd>
 813 <dt><strong><a name="item_error">Error</a></strong>
 814
 815 <dd>
 816 <pre>
 817         IV      (*Error)(pTHX_ PerlIO *f);</pre>
 818 </dd>
 819 <dd>
 820 <p>Return error indicator. <code>PerlIOBase_error()</code> is normally sufficient.</p>
 821 </dd>
 822 <dd>
 823 <p>Returns 1 if there is an error (usually when <a href="#item_perlio_f_error"><code>PERLIO_F_ERROR</code></a> is set,
 824 0 otherwise.</p>
 825 </dd>
 826 <dt><strong><a name="item_clearerr">Clearerr</a></strong>
 827
 828 <dd>
 829 <pre>
 830         void    (*Clearerr)(pTHX_ PerlIO *f);</pre>
 831 </dd>
 832 <dd>
 833 <p>Clear end-of-file and error indicators. Should call <code>PerlIOBase_clearerr()</code>
 834 to set the <code>PERLIO_F_XXXXX</code> flags, which may suffice.</p>
 835 </dd>
 836 <dt><strong><a name="item_setlinebuf">Setlinebuf</a></strong>
 837
 838 <dd>
 839 <pre>
 840         void    (*Setlinebuf)(pTHX_ PerlIO *f);</pre>
 841 </dd>
 842 <dd>
 843 <p>Mark the stream as line buffered. <code>PerlIOBase_setlinebuf()</code> sets the
 844 PERLIO_F_LINEBUF flag and is normally sufficient.</p>
 845 </dd>
 846 <dt><strong><a name="item_get_base">Get_base</a></strong>
 847
 848 <dd>
 849 <pre>
 850         STDCHAR *       (*Get_base)(pTHX_ PerlIO *f);</pre>
 851 </dd>
 852 <dd>
 853 <p>Allocate (if not already done so) the read buffer for this layer and
 854 return pointer to it. Return NULL on failure.</p>
 855 </dd>
 856 <dt><strong><a name="item_get_bufsiz">Get_bufsiz</a></strong>
 857
 858 <dd>
 859 <pre>
 860         Size_t  (*Get_bufsiz)(pTHX_ PerlIO *f);</pre>
 861 </dd>
 862 <dd>
 863 <p>Return the number of bytes that last <a href="#item_fill"><code>Fill()</code></a> put in the buffer.</p>
 864 </dd>
 865 <dt><strong><a name="item_get_ptr">Get_ptr</a></strong>
 866
 867 <dd>
 868 <pre>
 869         STDCHAR *       (*Get_ptr)(pTHX_ PerlIO *f);</pre>
 870 </dd>
 871 <dd>
 872 <p>Return the current read pointer relative to this layer's buffer.</p>
 873 </dd>
 874 <dt><strong><a name="item_get_cnt">Get_cnt</a></strong>
 875
 876 <dd>
 877 <pre>
 878         SSize_t (*Get_cnt)(pTHX_ PerlIO *f);</pre>
 879 </dd>
 880 <dd>
 881 <p>Return the number of bytes left to be read in the current buffer.</p>
 882 </dd>
 883 <dt><strong><a name="item_set_ptrcnt">Set_ptrcnt</a></strong>
 884
 885 <dd>
 886 <pre>
 887         void    (*Set_ptrcnt)(pTHX_ PerlIO *f,
 888                               STDCHAR *ptr, SSize_t cnt);</pre>
 889 </dd>
 890 <dd>
 891 <p>Adjust the read pointer and count of bytes to match <code>ptr</code> and/or <code>cnt</code>.
 892 The application (or layer above) must ensure they are consistent.
 893 (Checking is allowed by the paranoid.)</p>
 894 </dd>
 895 </dl>
 896 <p>
 897 </p>
 898 <h2><a name="utilities">Utilities</a></h2>
 899 <p>To ask for the next layer down use PerlIONext(PerlIO *f).</p>
 900 <p>To check that a PerlIO* is valid use PerlIOValid(PerlIO *f).  (All
 901 this does is really just to check that the pointer is non-NULL and
 902 that the pointer behind that is non-NULL.)</p>
 903 <p>PerlIOBase(PerlIO *f) returns the ``Base'' pointer, or in other words,
 904 the <code>PerlIOl*</code> pointer.</p>
 905 <p>PerlIOSelf(PerlIO* f, type) return the PerlIOBase cast to a type.</p>
 906 <p>Perl_PerlIO_or_Base(PerlIO* f, callback, base, failure, args) either
 907 calls the <em>callback</em> from the functions of the layer <em>f</em> (just by
 908 the name of the IO function, like ``Read'') with the <em>args</em>, or if
 909 there is no such callback, calls the <em>base</em> version of the callback
 910 with the same args, or if the f is invalid, set errno to EBADF and
 911 return <em>failure</em>.</p>
 912 <p>Perl_PerlIO_or_fail(PerlIO* f, callback, failure, args) either calls
 913 the <em>callback</em> of the functions of the layer <em>f</em> with the <em>args</em>,
 914 or if there is no such callback, set errno to EINVAL.  Or if the f is
 915 invalid, set errno to EBADF and return <em>failure</em>.</p>
 916 <p>Perl_PerlIO_or_Base_void(PerlIO* f, callback, base, args) either calls
 917 the <em>callback</em> of the functions of the layer <em>f</em> with the <em>args</em>,
 918 or if there is no such callback, calls the <em>base</em> version of the
 919 callback with the same args, or if the f is invalid, set errno to
 920 EBADF.</p>
 921 <p>Perl_PerlIO_or_fail_void(PerlIO* f, callback, args) either calls the
 922 <em>callback</em> of the functions of the layer <em>f</em> with the <em>args</em>, or if
 923 there is no such callback, set errno to EINVAL.  Or if the f is
 924 invalid, set errno to EBADF.</p>
 925 <p>
 926 </p>
 927 <h2><a name="implementing_perlio_layers">Implementing PerlIO Layers</a></h2>
 928 <p>If you find the implementation document unclear or not sufficient,
 929 look at the existing PerlIO layer implementations, which include:</p>
 930 <ul>
 931 <li><strong><a name="item_c_implementations">C implementations</a></strong>
 932
 933 <p>The <em>perlio.c</em> and <em>perliol.h</em> in the Perl core implement the
 934 ``unix'', ``perlio'', ``stdio'', ``crlf'', ``utf8'', ``byte'', ``raw'', ``pending''
 935 layers, and also the ``mmap'' and ``win32'' layers if applicable.
 936 (The ``win32'' is currently unfinished and unused, to see what is used
 937 instead in Win32, see <a href="file://C|\msysgit\mingw\html/lib/PerlIO/querying_the_layers_of_filehandles.html">Querying the layers of filehandles in the PerlIO manpage</a> .)</p>
 938 <p>PerlIO::encoding, PerlIO::scalar, PerlIO::via in the Perl core.</p>
 939 <p>PerlIO::gzip and APR::PerlIO (mod_perl 2.0) on CPAN.</p>
 940 </li>
 941 <li><strong><a name="item_perl_implementations">Perl implementations</a></strong>
 942
 943 <p>PerlIO::via::QuotedPrint in the Perl core and PerlIO::via::* on CPAN.</p>
 944 </li>
 945 </ul>
 946 <p>If you are creating a PerlIO layer, you may want to be lazy, in other
 947 words, implement only the methods that interest you.  The other methods
 948 you can either replace with the ``blank'' methods</p>
 949 <pre>
 950     PerlIOBase_noop_ok
 951     PerlIOBase_noop_fail</pre>
 952 <p>(which do nothing, and return zero and -1, respectively) or for
 953 certain methods you may assume a default behaviour by using a NULL
 954 method.  The Open method looks for help in the 'parent' layer.
 955 The following table summarizes the behaviour:</p>
 956 <pre>
 957     method      behaviour with NULL</pre>
 958 <pre>
 959     Clearerr    PerlIOBase_clearerr
 960     Close       PerlIOBase_close
 961     Dup         PerlIOBase_dup
 962     Eof         PerlIOBase_eof
 963     Error       PerlIOBase_error
 964     Fileno      PerlIOBase_fileno
 965     Fill        FAILURE
 966     Flush       SUCCESS
 967     Getarg      SUCCESS
 968     Get_base    FAILURE
 969     Get_bufsiz  FAILURE
 970     Get_cnt     FAILURE
 971     Get_ptr     FAILURE
 972     Open        INHERITED
 973     Popped      SUCCESS
 974     Pushed      SUCCESS
 975     Read        PerlIOBase_read
 976     Seek        FAILURE
 977     Set_cnt     FAILURE
 978     Set_ptrcnt  FAILURE
 979     Setlinebuf  PerlIOBase_setlinebuf
 980     Tell        FAILURE
 981     Unread      PerlIOBase_unread
 982     Write       FAILURE</pre>
 983 <pre>
 984  FAILURE        Set errno (to EINVAL in UNIXish, to LIB$_INVARG in VMS) and
 985                 return -1 (for numeric return values) or NULL (for pointers)
 986  INHERITED      Inherited from the layer below
 987  SUCCESS        Return 0 (for numeric return values) or a pointer</pre>
 988 <p>
 989 </p>
 990 <h2><a name="core_layers">Core Layers</a></h2>
 991 <p>The file <code>perlio.c</code> provides the following layers:</p>
 992 <dl>
 993 <dt><strong><a name="item__22unix_22">``unix''</a></strong>
 994
 995 <dd>
 996 <p>A basic non-buffered layer which calls Unix/POSIX <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_read"><code>read()</code></a>, <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_write"><code>write()</code></a>,
 997 <code>lseek()</code>, <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_close"><code>close()</code></a>. No buffering. Even on platforms that distinguish
 998 between O_TEXT and O_BINARY this layer is always O_BINARY.</p>
 999 </dd>
1000 </li>
1001 <dt><strong><a name="item__22perlio_22">``perlio''</a></strong>
1002
1003 <dd>
1004 <p>A very complete generic buffering layer which provides the whole of
1005 PerlIO API. It is also intended to be used as a ``base class'' for other
1006 layers. (For example its <a href="#item_read"><code>Read()</code></a> method is implemented in terms of
1007 the <a href="#item_get_cnt"><code>Get_cnt()</code></a>/<a href="#item_get_ptr"><code>Get_ptr()</code></a>/<a href="#item_set_ptrcnt"><code>Set_ptrcnt()</code></a> methods).</p>
1008 </dd>
1009 <dd>
1010 <p>``perlio'' over ``unix'' provides a complete replacement for stdio as seen
1011 via PerlIO API. This is the default for USE_PERLIO when system's stdio
1012 does not permit perl's ``fast gets'' access, and which do not
1013 distinguish between <code>O_TEXT</code> and <code>O_BINARY</code>.</p>
1014 </dd>
1015 </li>
1016 <dt><strong><a name="item__22stdio_22">``stdio''</a></strong>
1017
1018 <dd>
1019 <p>A layer which provides the PerlIO API via the layer scheme, but
1020 implements it by calling system's stdio. This is (currently) the default
1021 if system's stdio provides sufficient access to allow perl's ``fast gets''
1022 access and which do not distinguish between <code>O_TEXT</code> and <code>O_BINARY</code>.</p>
1023 </dd>
1024 </li>
1025 <dt><strong><a name="item__22crlf_22">``crlf''</a></strong>
1026
1027 <dd>
1028 <p>A layer derived using ``perlio'' as a base class. It provides Win32-like
1029 ``\n'' to CR,LF translation. Can either be applied above ``perlio'' or serve
1030 as the buffer layer itself. ``crlf'' over ``unix'' is the default if system
1031 distinguishes between <code>O_TEXT</code> and <code>O_BINARY</code> opens. (At some point
1032 ``unix'' will be replaced by a ``native'' Win32 IO layer on that platform,
1033 as Win32's read/write layer has various drawbacks.) The ``crlf'' layer is
1034 a reasonable model for a layer which transforms data in some way.</p>
1035 </dd>
1036 </li>
1037 <dt><strong><a name="item__22mmap_22">``mmap''</a></strong>
1038
1039 <dd>
1040 <p>If Configure detects <code>mmap()</code> functions this layer is provided (with
1041 ``perlio'' as a ``base'') which does ``read'' operations by mmap()ing the
1042 file. Performance improvement is marginal on modern systems, so it is
1043 mainly there as a proof of concept. It is likely to be unbundled from
1044 the core at some point. The ``mmap'' layer is a reasonable model for a
1045 minimalist ``derived'' layer.</p>
1046 </dd>
1047 </li>
1048 <dt><strong><a name="item__22pending_22">``pending''</a></strong>
1049
1050 <dd>
1051 <p>An ``internal'' derivative of ``perlio'' which can be used to provide
1052 <a href="#item_unread"><code>Unread()</code></a> function for layers which have no buffer or cannot be
1053 bothered.  (Basically this layer's <a href="#item_fill"><code>Fill()</code></a> pops itself off the stack
1054 and so resumes reading from layer below.)</p>
1055 </dd>
1056 </li>
1057 <dt><strong><a name="item__22raw_22">``raw''</a></strong>
1058
1059 <dd>
1060 <p>A dummy layer which never exists on the layer stack. Instead when
1061 ``pushed'' it actually pops the stack removing itself, it then calls
1062 Binmode function table entry on all the layers in the stack - normally
1063 this (via PerlIOBase_binmode) removes any layers which do not have
1064 <a href="#item_perlio_k_raw"><code>PERLIO_K_RAW</code></a> bit set. Layers can modify that behaviour by defining
1065 their own Binmode entry.</p>
1066 </dd>
1067 </li>
1068 <dt><strong><a name="item__22utf8_22">``utf8''</a></strong>
1069
1070 <dd>
1071 <p>Another dummy layer. When pushed it pops itself and sets the
1072 <a href="#item_perlio_f_utf8"><code>PERLIO_F_UTF8</code></a> flag on the layer which was (and now is once more)
1073 the top of the stack.</p>
1074 </dd>
1075 </li>
1076 </dl>
1077 <p>In addition <em>perlio.c</em> also provides a number of <code>PerlIOBase_xxxx()</code>
1078 functions which are intended to be used in the table slots of classes
1079 which do not need to do anything special for a particular method.</p>
1080 <p>
1081 </p>
1082 <h2><a name="extension_layers">Extension Layers</a></h2>
1083 <p>Layers can made available by extension modules. When an unknown layer
1084 is encountered the PerlIO code will perform the equivalent of :</p>
1085 <pre>
1086    use PerlIO 'layer';</pre>
1087 <p>Where <em>layer</em> is the unknown layer. <em>PerlIO.pm</em> will then attempt to:</p>
1088 <pre>
1089    require PerlIO::layer;</pre>
1090 <p>If after that process the layer is still not defined then the <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a>
1091 will fail.</p>
1092 <p>The following extension layers are bundled with perl:</p>
1093 <dl>
1094 <dt><strong><a name="item__22_3aencoding_22">``:encoding''</a></strong>
1095
1096 <dd>
1097 <pre>
1098    use Encoding;</pre>
1099 </dd>
1100 <dd>
1101 <p>makes this layer available, although <em>PerlIO.pm</em> ``knows'' where to
1102 find it.  It is an example of a layer which takes an argument as it is
1103 called thus:</p>
1104 </dd>
1105 <dd>
1106 <pre>
1107    open( $fh, &quot;&lt;:encoding(iso-8859-7)&quot;, $pathname );</pre>
1108 </dd>
1109 <dt><strong><a name="item__22_3ascalar_22">``:scalar''</a></strong>
1110
1111 <dd>
1112 <p>Provides support for reading data from and writing data to a scalar.</p>
1113 </dd>
1114 <dd>
1115 <pre>
1116    open( $fh, &quot;+&lt;:scalar&quot;, \$scalar );</pre>
1117 </dd>
1118 <dd>
1119 <p>When a handle is so opened, then reads get bytes from the string value
1120 of <em>$scalar</em>, and writes change the value. In both cases the position
1121 in <em>$scalar</em> starts as zero but can be altered via <code>seek</code>, and
1122 determined via <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_tell"><code>tell</code></a>.</p>
1123 </dd>
1124 <dd>
1125 <p>Please note that this layer is implied when calling <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open()</code></a> thus:</p>
1126 </dd>
1127 <dd>
1128 <pre>
1129    open( $fh, &quot;+&lt;&quot;, \$scalar );</pre>
1130 </dd>
1131 </li>
1132 <dt><strong><a name="item__22_3avia_22">``:via''</a></strong>
1133
1134 <dd>
1135 <p>Provided to allow layers to be implemented as Perl code.  For instance:</p>
1136 </dd>
1137 <dd>
1138 <pre>
1139    use PerlIO::via::StripHTML;
1140    open( my $fh, &quot;&lt;:via(StripHTML)&quot;, &quot;index.html&quot; );</pre>
1141 </dd>
1142 <dd>
1143 <p>See <a href="file://C|\msysgit\mingw\html/lib/PerlIO/via.html">the PerlIO::via manpage</a> for details.</p>
1144 </dd>
1145 </li>
1146 </dl>
1147 <p>
1148 </p>
1149 <hr />
1150 <h1><a name="todo">TODO</a></h1>
1151 <p>Things that need to be done to improve this document.</p>
1152 <ul>
1153 <li>
1154 <p>Explain how to make a valid fh without going through open()(i.e. apply
1155 a layer). For example if the file is not opened through perl, but we
1156 want to get back a fh, like it was opened by Perl.</p>
1157 <p>How PerlIO_apply_layera fits in, where its docs, was it made public?</p>
1158 <p>Currently the example could be something like this:</p>
1159 <pre>
1160   PerlIO *foo_to_PerlIO(pTHX_ char *mode, ...)
1161   {
1162       char *mode; /* &quot;w&quot;, &quot;r&quot;, etc */
1163       const char *layers = &quot;:APR&quot;; /* the layer name */
1164       PerlIO *f = PerlIO_allocate(aTHX);
1165       if (!f) {
1166           return NULL;
1167       }</pre>
1168 <pre>
1169       PerlIO_apply_layers(aTHX_ f, mode, layers);</pre>
1170 <pre>
1171       if (f) {
1172           PerlIOAPR *st = PerlIOSelf(f, PerlIOAPR);
1173           /* fill in the st struct, as in _open() */
1174           st-&gt;file = file;
1175           PerlIOBase(f)-&gt;flags |= PERLIO_F_OPEN;</pre>
1176 <pre>
1177           return f;
1178       }
1179       return NULL;
1180   }</pre>
1181 </li>
1182 <li>
1183 <p>fix/add the documentation in places marked as XXX.</p>
1184 </li>
1185 <li>
1186 <p>The handling of errors by the layer is not specified. e.g. when $!
1187 should be set explicitly, when the error handling should be just
1188 delegated to the top layer.</p>
1189 <p>Probably give some hints on using <code>SETERRNO()</code> or pointers to where they
1190 can be found.</p>
1191 </li>
1192 <li>
1193 <p>I think it would help to give some concrete examples to make it easier
1194 to understand the API. Of course I agree that the API has to be
1195 concise, but since there is no second document that is more of a
1196 guide, I think that it'd make it easier to start with the doc which is
1197 an API, but has examples in it in places where things are unclear, to
1198 a person who is not a PerlIO guru (yet).</p>
1199 </li>
1200 </ul>
1201 <table border="0" width="100%" cellspacing="0" cellpadding="3">
1202 <tr><td class="block" style="background-color: #cccccc" valign="middle">
1203 <big><strong><span class="block">&nbsp;perliol - C API for Perl's implementation of IO in Layers.</span></strong></big>
1204 </td></tr>
1205 </table>
1206
1207 </body>
1208
1209 </html>