mingw/html/lib/Tie/File.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>Tie::File - Access the lines of a disk file via a Perl array</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;Tie::File - Access the lines of a disk file via a Perl array</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="#recsep"><code>recsep</code></a></li>
  28                 <li><a href="#autochomp"><code>autochomp</code></a></li>
  29                 <li><a href="#mode"><code>mode</code></a></li>
  30                 <li><a href="#memory"><code>memory</code></a></li>
  31                 <li><a href="#dw_size"><code>dw_size</code></a></li>
  32                 <li><a href="#option_format">Option Format</a></li>
  33         </ul>
  34
  35         <li><a href="#public_methods">Public Methods</a></li>
  36         <ul>
  37
  38                 <li><a href="#flock"><code>flock</code></a></li>
  39                 <li><a href="#autochomp"><code>autochomp</code></a></li>
  40                 <li><a href="#defer__flush__discard__and_autodefer"><code>defer</code>, <code>flush</code>, <code>discard</code>, and <code>autodefer</code></a></li>
  41                 <li><a href="#offset"><code>offset</code></a></li>
  42         </ul>
  43
  44         <li><a href="#tying_to_an_alreadyopened_filehandle">Tying to an already-opened filehandle</a></li>
  45         <li><a href="#deferred_writing">Deferred Writing</a></li>
  46         <ul>
  47
  48                 <li><a href="#autodeferring">Autodeferring</a></li>
  49         </ul>
  50
  51         <li><a href="#concurrent_access_to_files">CONCURRENT ACCESS TO FILES</a></li>
  52         <li><a href="#caveats">CAVEATS</a></li>
  53         <li><a href="#subclassing">SUBCLASSING</a></li>
  54         <li><a href="#what_about_db_file">WHAT ABOUT <code>DB_File</code>?</a></li>
  55         <li><a href="#author">AUTHOR</a></li>
  56         <li><a href="#license">LICENSE</a></li>
  57         <li><a href="#warranty">WARRANTY</a></li>
  58         <li><a href="#thanks">THANKS</a></li>
  59         <li><a href="#todo">TODO</a></li>
  60 </ul>
  61 <!-- INDEX END -->
  62
  63 <hr />
  64 <p>
  65 </p>
  66 <h1><a name="name">NAME</a></h1>
  67 <p>Tie::File - Access the lines of a disk file via a Perl array</p>
  68 <p>
  69 </p>
  70 <hr />
  71 <h1><a name="synopsis">SYNOPSIS</a></h1>
  72 <pre>
  73         # This file documents Tie::File version 0.97
  74         use Tie::File;</pre>
  75 <pre>
  76         tie @array, 'Tie::File', filename or die ...;</pre>
  77 <pre>
  78         $array[13] = 'blah';     # line 13 of the file is now 'blah'
  79         print $array[42];        # display line 42 of the file</pre>
  80 <pre>
  81         $n_recs = @array;        # how many records are in the file?
  82         $#array -= 2;            # chop two records off the end</pre>
  83 <pre>
  84         for (@array) {
  85           s/PERL/Perl/g;         # Replace PERL with Perl everywhere in the file
  86         }</pre>
  87 <pre>
  88         # These are just like regular push, pop, unshift, shift, and splice
  89         # Except that they modify the file in the way you would expect</pre>
  90 <pre>
  91         push @array, new recs...;
  92         my $r1 = pop @array;
  93         unshift @array, new recs...;
  94         my $r2 = shift @array;
  95         @old_recs = splice @array, 3, 7, new recs...;</pre>
  96 <pre>
  97         untie @array;            # all finished</pre>
  98 <p>
  99 </p>
 100 <hr />
 101 <h1><a name="description">DESCRIPTION</a></h1>
 102 <p><code>Tie::File</code> represents a regular text file as a Perl array.  Each
 103 element in the array corresponds to a record in the file.  The first
 104 line of the file is element 0 of the array; the second line is element
 105 1, and so on.</p>
 106 <p>The file is <em>not</em> loaded into memory, so this will work even for
 107 gigantic files.</p>
 108 <p>Changes to the array are reflected in the file immediately.</p>
 109 <p>Lazy people and beginners may now stop reading the manual.</p>
 110 <p>
 111 </p>
 112 <h2><a name="recsep"><code>recsep</code></a></h2>
 113 <p>What is a 'record'?  By default, the meaning is the same as for the
 114 <code>&lt;...&gt;</code> operator: It's a string terminated by <a href="file://C|\msysgit\mingw\html/pod/perlvar.html#item___"><code>$/</code></a>, which is
 115 probably <code>&quot;\n&quot;</code>.  (Minor exception: on DOS and Win32 systems, a
 116 'record' is a string terminated by <code>&quot;\r\n&quot;</code>.)  You may change the
 117 definition of ``record'' by supplying the <code>recsep</code> option in the <code>tie</code>
 118 call:</p>
 119 <pre>
 120         tie @array, 'Tie::File', $file, recsep =&gt; 'es';</pre>
 121 <p>This says that records are delimited by the string <code>es</code>.  If the file
 122 contained the following data:</p>
 123 <pre>
 124         Curse these pesky flies!\n</pre>
 125 <p>then the <code>@array</code> would appear to have four elements:</p>
 126 <pre>
 127         &quot;Curse th&quot;
 128         &quot;e p&quot;
 129         &quot;ky fli&quot;
 130         &quot;!\n&quot;</pre>
 131 <p>An undefined value is not permitted as a record separator.  Perl's
 132 special ``paragraph mode'' semantics (&agrave; la <a href="file://C|\msysgit\mingw\html/pod/perlvar.html#item___"><code>$/ = &quot;&quot;</code></a>) are not
 133 emulated.</p>
 134 <p>Records read from the tied array do not have the record separator
 135 string on the end; this is to allow</p>
 136 <pre>
 137         $array[17] .= &quot;extra&quot;;</pre>
 138 <p>to work as expected.</p>
 139 <p>(See <a href="#autochomp">autochomp</a>, below.)  Records stored into the array will have
 140 the record separator string appended before they are written to the
 141 file, if they don't have one already.  For example, if the record
 142 separator string is <code>&quot;\n&quot;</code>, then the following two lines do exactly
 143 the same thing:</p>
 144 <pre>
 145         $array[17] = &quot;Cherry pie&quot;;
 146         $array[17] = &quot;Cherry pie\n&quot;;</pre>
 147 <p>The result is that the contents of line 17 of the file will be
 148 replaced with ``Cherry pie''; a newline character will separate line 17
 149 from line 18.  This means that this code will do nothing:</p>
 150 <pre>
 151         chomp $array[17];</pre>
 152 <p>Because the <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_chomp"><code>chomp</code></a>ed value will have the separator reattached when
 153 it is written back to the file.  There is no way to create a file
 154 whose trailing record separator string is missing.</p>
 155 <p>Inserting records that <em>contain</em> the record separator string is not
 156 supported by this module.  It will probably produce a reasonable
 157 result, but what this result will be may change in a future version.
 158 Use 'splice' to insert records or to replace one record with several.</p>
 159 <p>
 160 </p>
 161 <h2><a name="autochomp"><code>autochomp</code></a></h2>
 162 <p>Normally, array elements have the record separator removed, so that if
 163 the file contains the text</p>
 164 <pre>
 165         Gold
 166         Frankincense
 167         Myrrh</pre>
 168 <p>the tied array will appear to contain <code>(&quot;Gold&quot;, &quot;Frankincense&quot;,
 169 &quot;Myrrh&quot;)</code>.  If you set <code>autochomp</code> to a false value, the record
 170 separator will not be removed.  If the file above was tied with</p>
 171 <pre>
 172         tie @gifts, &quot;Tie::File&quot;, $gifts, autochomp =&gt; 0;</pre>
 173 <p>then the array <code>@gifts</code> would appear to contain <code>(&quot;Gold\n&quot;,
 174 &quot;Frankincense\n&quot;, &quot;Myrrh\n&quot;)</code>, or (on Win32 systems) <code>(&quot;Gold\r\n&quot;,
 175 &quot;Frankincense\r\n&quot;, &quot;Myrrh\r\n&quot;)</code>.</p>
 176 <p>
 177 </p>
 178 <h2><a name="mode"><code>mode</code></a></h2>
 179 <p>Normally, the specified file will be opened for read and write access,
 180 and will be created if it does not exist.  (That is, the flags
 181 <code>O_RDWR | O_CREAT</code> are supplied in the <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a> call.)  If you want to
 182 change this, you may supply alternative flags in the <code>mode</code> option.
 183 See <a href="file://C|\msysgit\mingw\html/lib/Fcntl.html">the Fcntl manpage</a> for a listing of available flags.
 184 For example:</p>
 185 <pre>
 186         # open the file if it exists, but fail if it does not exist
 187         use Fcntl 'O_RDWR';
 188         tie @array, 'Tie::File', $file, mode =&gt; O_RDWR;</pre>
 189 <pre>
 190         # create the file if it does not exist
 191         use Fcntl 'O_RDWR', 'O_CREAT';
 192         tie @array, 'Tie::File', $file, mode =&gt; O_RDWR | O_CREAT;</pre>
 193 <pre>
 194         # open an existing file in read-only mode
 195         use Fcntl 'O_RDONLY';
 196         tie @array, 'Tie::File', $file, mode =&gt; O_RDONLY;</pre>
 197 <p>Opening the data file in write-only or append mode is not supported.</p>
 198 <p>
 199 </p>
 200 <h2><a name="memory"><code>memory</code></a></h2>
 201 <p>This is an upper limit on the amount of memory that <code>Tie::File</code> will
 202 consume at any time while managing the file.  This is used for two
 203 things: managing the <em>read cache</em> and managing the <em>deferred write
 204 buffer</em>.</p>
 205 <p>Records read in from the file are cached, to avoid having to re-read
 206 them repeatedly.  If you read the same record twice, the first time it
 207 will be stored in memory, and the second time it will be fetched from
 208 the <em>read cache</em>.  The amount of data in the read cache will not
 209 exceed the value you specified for <code>memory</code>.  If <code>Tie::File</code> wants
 210 to cache a new record, but the read cache is full, it will make room
 211 by expiring the least-recently visited records from the read cache.</p>
 212 <p>The default memory limit is 2Mib.  You can adjust the maximum read
 213 cache size by supplying the <code>memory</code> option.  The argument is the
 214 desired cache size, in bytes.</p>
 215 <pre>
 216         # I have a lot of memory, so use a large cache to speed up access
 217         tie @array, 'Tie::File', $file, memory =&gt; 20_000_000;</pre>
 218 <p>Setting the memory limit to 0 will inhibit caching; records will be
 219 fetched from disk every time you examine them.</p>
 220 <p>The <code>memory</code> value is not an absolute or exact limit on the memory
 221 used.  <code>Tie::File</code> objects contains some structures besides the read
 222 cache and the deferred write buffer, whose sizes are not charged
 223 against <code>memory</code>.</p>
 224 <p>The cache itself consumes about 310 bytes per cached record, so if
 225 your file has many short records, you may want to decrease the cache
 226 memory limit, or else the cache overhead may exceed the size of the
 227 cached data.</p>
 228 <p>
 229 </p>
 230 <h2><a name="dw_size"><code>dw_size</code></a></h2>
 231 <p>(This is an advanced feature.  Skip this section on first reading.)</p>
 232 <p>If you use deferred writing (See <a href="#deferred_writing">Deferred Writing</a>, below) then
 233 data you write into the array will not be written directly to the
 234 file; instead, it will be saved in the <em>deferred write buffer</em> to be
 235 written out later.  Data in the deferred write buffer is also charged
 236 against the memory limit you set with the <code>memory</code> option.</p>
 237 <p>You may set the <code>dw_size</code> option to limit the amount of data that can
 238 be saved in the deferred write buffer.  This limit may not exceed the
 239 total memory limit.  For example, if you set <code>dw_size</code> to 1000 and
 240 <code>memory</code> to 2500, that means that no more than 1000 bytes of deferred
 241 writes will be saved up.  The space available for the read cache will
 242 vary, but it will always be at least 1500 bytes (if the deferred write
 243 buffer is full) and it could grow as large as 2500 bytes (if the
 244 deferred write buffer is empty.)</p>
 245 <p>If you don't specify a <code>dw_size</code>, it defaults to the entire memory
 246 limit.</p>
 247 <p>
 248 </p>
 249 <h2><a name="option_format">Option Format</a></h2>
 250 <p><code>-mode</code> is a synonym for <code>mode</code>.  <code>-recsep</code> is a synonym for
 251 <code>recsep</code>.  <code>-memory</code> is a synonym for <code>memory</code>.  You get the
 252 idea.</p>
 253 <p>
 254 </p>
 255 <hr />
 256 <h1><a name="public_methods">Public Methods</a></h1>
 257 <p>The <code>tie</code> call returns an object, say <code>$o</code>.  You may call</p>
 258 <pre>
 259         $rec = $o-&gt;FETCH($n);
 260         $o-&gt;STORE($n, $rec);</pre>
 261 <p>to fetch or store the record at line <code>$n</code>, respectively; similarly
 262 the other tied array methods.  (See <a href="file://C|\msysgit\mingw\html/pod/perltie.html">the perltie manpage</a> for details.)  You may
 263 also call the following methods on this object:</p>
 264 <p>
 265 </p>
 266 <h2><a name="flock"><code>flock</code></a></h2>
 267 <pre>
 268         $o-&gt;flock(MODE)</pre>
 269 <p>will lock the tied file.  <code>MODE</code> has the same meaning as the second
 270 argument to the Perl built-in <code>flock</code> function; for example
 271 <code>LOCK_SH</code> or <code>LOCK_EX | LOCK_NB</code>.  (These constants are provided by
 272 the <code>use Fcntl ':flock'</code> declaration.)</p>
 273 <p><code>MODE</code> is optional; the default is <code>LOCK_EX</code>.</p>
 274 <p><code>Tie::File</code> maintains an internal table of the byte offset of each
 275 record it has seen in the file.</p>
 276 <p>When you use <code>flock</code> to lock the file, <code>Tie::File</code> assumes that the
 277 read cache is no longer trustworthy, because another process might
 278 have modified the file since the last time it was read.  Therefore, a
 279 successful call to <code>flock</code> discards the contents of the read cache
 280 and the internal record offset table.</p>
 281 <p><code>Tie::File</code> promises that the following sequence of operations will
 282 be safe:</p>
 283 <pre>
 284         my $o = tie @array, &quot;Tie::File&quot;, $filename;
 285         $o-&gt;flock;</pre>
 286 <p>In particular, <code>Tie::File</code> will <em>not</em> read or write the file during
 287 the <code>tie</code> call.  (Exception: Using <code>mode =&gt; O_TRUNC</code> will, of
 288 course, erase the file during the <code>tie</code> call.  If you want to do this
 289 safely, then open the file without <code>O_TRUNC</code>, lock the file, and use
 290 <code>@array = ()</code>.)</p>
 291 <p>The best way to unlock a file is to discard the object and untie the
 292 array.  It is probably unsafe to unlock the file without also untying
 293 it, because if you do, changes may remain unwritten inside the object.
 294 That is why there is no shortcut for unlocking.  If you really want to
 295 unlock the file prematurely, you know what to do; if you don't know
 296 what to do, then don't do it.</p>
 297 <p>All the usual warnings about file locking apply here.  In particular,
 298 note that file locking in Perl is <strong>advisory</strong>, which means that
 299 holding a lock will not prevent anyone else from reading, writing, or
 300 erasing the file; it only prevents them from getting another lock at
 301 the same time.  Locks are analogous to green traffic lights: If you
 302 have a green light, that does not prevent the idiot coming the other
 303 way from plowing into you sideways; it merely guarantees to you that
 304 the idiot does not also have a green light at the same time.</p>
 305 <p>
 306 </p>
 307 <h2><a name="autochomp"><code>autochomp</code></a></h2>
 308 <pre>
 309         my $old_value = $o-&gt;autochomp(0);    # disable autochomp option
 310         my $old_value = $o-&gt;autochomp(1);    #  enable autochomp option</pre>
 311 <pre>
 312         my $ac = $o-&gt;autochomp();   # recover current value</pre>
 313 <p>See <a href="#autochomp">autochomp</a>, above.</p>
 314 <p>
 315 </p>
 316 <h2><a name="defer__flush__discard__and_autodefer"><code>defer</code>, <code>flush</code>, <code>discard</code>, and <code>autodefer</code></a></h2>
 317 <p>See <a href="#deferred_writing">Deferred Writing</a>, below.</p>
 318 <p>
 319 </p>
 320 <h2><a name="offset"><code>offset</code></a></h2>
 321 <pre>
 322         $off = $o-&gt;offset($n);</pre>
 323 <p>This method returns the byte offset of the start of the <code>$n</code>th record
 324 in the file.  If there is no such record, it returns an undefined
 325 value.</p>
 326 <p>
 327 </p>
 328 <hr />
 329 <h1><a name="tying_to_an_alreadyopened_filehandle">Tying to an already-opened filehandle</a></h1>
 330 <p>If <code>$fh</code> is a filehandle, such as is returned by <code>IO::File</code> or one
 331 of the other <code>IO</code> modules, you may use:</p>
 332 <pre>
 333         tie @array, 'Tie::File', $fh, ...;</pre>
 334 <p>Similarly if you opened that handle <code>FH</code> with regular <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_open"><code>open</code></a> or
 335 <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_sysopen"><code>sysopen</code></a>, you may use:</p>
 336 <pre>
 337         tie @array, 'Tie::File', \*FH, ...;</pre>
 338 <p>Handles that were opened write-only won't work.  Handles that were
 339 opened read-only will work as long as you don't try to modify the
 340 array.  Handles must be attached to seekable sources of data---that
 341 means no pipes or sockets.  If <code>Tie::File</code> can detect that you
 342 supplied a non-seekable handle, the <code>tie</code> call will throw an
 343 exception.  (On Unix systems, it can detect this.)</p>
 344 <p>Note that Tie::File will only close any filehandles that it opened
 345 internally.  If you passed it a filehandle as above, you ``own'' the
 346 filehandle, and are responsible for closing it after you have untied
 347 the @array.</p>
 348 <p>
 349 </p>
 350 <hr />
 351 <h1><a name="deferred_writing">Deferred Writing</a></h1>
 352 <p>(This is an advanced feature.  Skip this section on first reading.)</p>
 353 <p>Normally, modifying a <code>Tie::File</code> array writes to the underlying file
 354 immediately.  Every assignment like <code>$a[3] = ...</code> rewrites as much of
 355 the file as is necessary; typically, everything from line 3 through
 356 the end will need to be rewritten.  This is the simplest and most
 357 transparent behavior.  Performance even for large files is reasonably
 358 good.</p>
 359 <p>However, under some circumstances, this behavior may be excessively
 360 slow.  For example, suppose you have a million-record file, and you
 361 want to do:</p>
 362 <pre>
 363         for (@FILE) {
 364           $_ = &quot;&gt; $_&quot;;
 365         }</pre>
 366 <p>The first time through the loop, you will rewrite the entire file,
 367 from line 0 through the end.  The second time through the loop, you
 368 will rewrite the entire file from line 1 through the end.  The third
 369 time through the loop, you will rewrite the entire file from line 2 to
 370 the end.  And so on.</p>
 371 <p>If the performance in such cases is unacceptable, you may defer the
 372 actual writing, and then have it done all at once.  The following loop
 373 will perform much better for large files:</p>
 374 <pre>
 375         (tied @a)-&gt;defer;
 376         for (@a) {
 377           $_ = &quot;&gt; $_&quot;;
 378         }
 379         (tied @a)-&gt;flush;</pre>
 380 <p>If <code>Tie::File</code>'s memory limit is large enough, all the writing will
 381 done in memory.  Then, when you call <code>-&gt;flush</code>, the entire file
 382 will be rewritten in a single pass.</p>
 383 <p>(Actually, the preceding discussion is something of a fib.  You don't
 384 need to enable deferred writing to get good performance for this
 385 common case, because <code>Tie::File</code> will do it for you automatically
 386 unless you specifically tell it not to.  See <a href="#autodeferring">autodeferring</a>,
 387 below.)</p>
 388 <p>Calling <code>-&gt;flush</code> returns the array to immediate-write mode.  If
 389 you wish to discard the deferred writes, you may call <code>-&gt;discard</code>
 390 instead of <code>-&gt;flush</code>.  Note that in some cases, some of the data
 391 will have been written already, and it will be too late for
 392 <code>-&gt;discard</code> to discard all the changes.  Support for
 393 <code>-&gt;discard</code> may be withdrawn in a future version of <code>Tie::File</code>.</p>
 394 <p>Deferred writes are cached in memory up to the limit specified by the
 395 <code>dw_size</code> option (see above).  If the deferred-write buffer is full
 396 and you try to write still more deferred data, the buffer will be
 397 flushed.  All buffered data will be written immediately, the buffer
 398 will be emptied, and the now-empty space will be used for future
 399 deferred writes.</p>
 400 <p>If the deferred-write buffer isn't yet full, but the total size of the
 401 buffer and the read cache would exceed the <code>memory</code> limit, the oldest
 402 records will be expired from the read cache until the total size is
 403 under the limit.</p>
 404 <p><code>push</code>, <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_pop"><code>pop</code></a>, <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_shift"><code>shift</code></a>, <code>unshift</code>, and <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_splice"><code>splice</code></a> cannot be
 405 deferred.  When you perform one of these operations, any deferred data
 406 is written to the file and the operation is performed immediately.
 407 This may change in a future version.</p>
 408 <p>If you resize the array with deferred writing enabled, the file will
 409 be resized immediately, but deferred records will not be written.
 410 This has a surprising consequence: <code>@a = (...)</code> erases the file
 411 immediately, but the writing of the actual data is deferred.  This
 412 might be a bug.  If it is a bug, it will be fixed in a future version.</p>
 413 <p>
 414 </p>
 415 <h2><a name="autodeferring">Autodeferring</a></h2>
 416 <p><code>Tie::File</code> tries to guess when deferred writing might be helpful,
 417 and to turn it on and off automatically.</p>
 418 <pre>
 419         for (@a) {
 420           $_ = &quot;&gt; $_&quot;;
 421         }</pre>
 422 <p>In this example, only the first two assignments will be done
 423 immediately; after this, all the changes to the file will be deferred
 424 up to the user-specified memory limit.</p>
 425 <p>You should usually be able to ignore this and just use the module
 426 without thinking about deferring.  However, special applications may
 427 require fine control over which writes are deferred, or may require
 428 that all writes be immediate.  To disable the autodeferment feature,
 429 use</p>
 430 <pre>
 431         (tied @o)-&gt;autodefer(0);</pre>
 432 <p>or</p>
 433 <pre>
 434         tie @array, 'Tie::File', $file, autodefer =&gt; 0;</pre>
 435 <p>Similarly, <code>-&gt;autodefer(1)</code> re-enables autodeferment, and
 436 <code>-&gt;autodefer()</code> recovers the current value of the autodefer setting.</p>
 437 <p>
 438 </p>
 439 <hr />
 440 <h1><a name="concurrent_access_to_files">CONCURRENT ACCESS TO FILES</a></h1>
 441 <p>Caching and deferred writing are inappropriate if you want the same
 442 file to be accessed simultaneously from more than one process.  Other
 443 optimizations performed internally by this module are also
 444 incompatible with concurrent access.  A future version of this module will
 445 support a <code>concurrent =&gt; 1</code> option that enables safe concurrent access.</p>
 446 <p>Previous versions of this documentation suggested using <code>memory
 447 =E&lt;gt&gt; 0</code> for safe concurrent access.  This was mistaken.  Tie::File
 448 will not support safe concurrent access before version 0.98.</p>
 449 <p>
 450 </p>
 451 <hr />
 452 <h1><a name="caveats">CAVEATS</a></h1>
 453 <p>(That's Latin for 'warnings'.)</p>
 454 <ul>
 455 <li>
 456 <p>Reasonable effort was made to make this module efficient.  Nevertheless,
 457 changing the size of a record in the middle of a large file will
 458 always be fairly slow, because everything after the new record must be
 459 moved.</p>
 460 </li>
 461 <li>
 462 <p>The behavior of tied arrays is not precisely the same as for regular
 463 arrays.  For example:</p>
 464 <pre>
 465         # This DOES print &quot;How unusual!&quot;
 466         undef $a[10];  print &quot;How unusual!\n&quot; if defined $a[10];</pre>
 467 <p><a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_undef"><code>undef</code></a>-ing a <code>Tie::File</code> array element just blanks out the
 468 corresponding record in the file.  When you read it back again, you'll
 469 get the empty string, so the supposedly-<a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_undef"><code>undef</code></a>'ed value will be
 470 defined.  Similarly, if you have <code>autochomp</code> disabled, then</p>
 471 <pre>
 472         # This DOES print &quot;How unusual!&quot; if 'autochomp' is disabled
 473         undef $a[10];
 474         print &quot;How unusual!\n&quot; if $a[10];</pre>
 475 <p>Because when <code>autochomp</code> is disabled, <code>$a[10]</code> will read back as
 476 <code>&quot;\n&quot;</code> (or whatever the record separator string is.)</p>
 477 <p>There are other minor differences, particularly regarding <code>exists</code>
 478 and <code>delete</code>, but in general, the correspondence is extremely close.</p>
 479 </li>
 480 <li>
 481 <p>I have supposed that since this module is concerned with file I/O,
 482 almost all normal use of it will be heavily I/O bound.  This means
 483 that the time to maintain complicated data structures inside the
 484 module will be dominated by the time to actually perform the I/O.
 485 When there was an opportunity to spend CPU time to avoid doing I/O, I
 486 usually tried to take it.</p>
 487 </li>
 488 <li>
 489 <p>You might be tempted to think that deferred writing is like
 490 transactions, with <code>flush</code> as <code>commit</code> and <code>discard</code> as
 491 <code>rollback</code>, but it isn't, so don't.</p>
 492 </li>
 493 <li>
 494 <p>There is a large memory overhead for each record offset and for each
 495 cache entry: about 310 bytes per cached data record, and about 21 bytes per offset table entry.</p>
 496 <p>The per-record overhead will limit the maximum number of records you
 497 can access per file. Note that <em>accessing</em> the length of the array
 498 via <code>$x = scalar @tied_file</code> accesses <strong>all</strong> records and stores their
 499 offsets.  The same for <code>foreach (@tied_file)</code>, even if you exit the
 500 loop early.</p>
 501 </li>
 502 </ul>
 503 <p>
 504 </p>
 505 <hr />
 506 <h1><a name="subclassing">SUBCLASSING</a></h1>
 507 <p>This version promises absolutely nothing about the internals, which
 508 may change without notice.  A future version of the module will have a
 509 well-defined and stable subclassing API.</p>
 510 <p>
 511 </p>
 512 <hr />
 513 <h1><a name="what_about_db_file">WHAT ABOUT <code>DB_File</code>?</a></h1>
 514 <p>People sometimes point out that <a href="file://C|\msysgit\mingw\html/ext/DB_File/DB_File.html">the DB_File manpage</a> will do something similar,
 515 and ask why <code>Tie::File</code> module is necessary.</p>
 516 <p>There are a number of reasons that you might prefer <code>Tie::File</code>.
 517 A list is available at <code>http://perl.plover.com/TieFile/why-not-DB_File</code>.</p>
 518 <p>
 519 </p>
 520 <hr />
 521 <h1><a name="author">AUTHOR</a></h1>
 522 <p>Mark Jason Dominus</p>
 523 <p>To contact the author, send email to: <code>mjd-perl-tiefile+@plover.com</code></p>
 524 <p>To receive an announcement whenever a new version of this module is
 525 released, send a blank email message to
 526 <code>mjd-perl-tiefile-subscribe@plover.com</code>.</p>
 527 <p>The most recent version of this module, including documentation and
 528 any news of importance, will be available at</p>
 529 <pre>
 530         <a href="http://perl.plover.com/TieFile/">http://perl.plover.com/TieFile/</a></pre>
 531 <p>
 532 </p>
 533 <hr />
 534 <h1><a name="license">LICENSE</a></h1>
 535 <p><code>Tie::File</code> version 0.97 is copyright (C) 2003 Mark Jason Dominus.</p>
 536 <p>This library is free software; you may redistribute it and/or modify
 537 it under the same terms as Perl itself.</p>
 538 <p>These terms are your choice of any of (1) the Perl Artistic Licence,
 539 or (2) version 2 of the GNU General Public License as published by the
 540 Free Software Foundation, or (3) any later version of the GNU General
 541 Public License.</p>
 542 <p>This library is distributed in the hope that it will be useful,
 543 but WITHOUT ANY WARRANTY; without even the implied warranty of
 544 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 545 GNU General Public License for more details.</p>
 546 <p>You should have received a copy of the GNU General Public License
 547 along with this library program; it should be in the file <code>COPYING</code>.
 548 If not, write to the Free Software Foundation, Inc., 59 Temple Place,
 549 Suite 330, Boston, MA 02111 USA</p>
 550 <p>For licensing inquiries, contact the author at:</p>
 551 <pre>
 552         Mark Jason Dominus
 553         255 S. Warnock St.
 554         Philadelphia, PA 19107</pre>
 555 <p>
 556 </p>
 557 <hr />
 558 <h1><a name="warranty">WARRANTY</a></h1>
 559 <p><code>Tie::File</code> version 0.97 comes with ABSOLUTELY NO WARRANTY.
 560 For details, see the license.</p>
 561 <p>
 562 </p>
 563 <hr />
 564 <h1><a name="thanks">THANKS</a></h1>
 565 <p>Gigantic thanks to Jarkko Hietaniemi, for agreeing to put this in the
 566 core when I hadn't written it yet, and for generally being helpful,
 567 supportive, and competent.  (Usually the rule is ``choose any one.'')
 568 Also big thanks to Abhijit Menon-Sen for all of the same things.</p>
 569 <p>Special thanks to Craig Berry and Peter Prymmer (for VMS portability
 570 help), Randy Kobes (for Win32 portability help), Clinton Pierce and
 571 Autrijus Tang (for heroic eleventh-hour Win32 testing above and beyond
 572 the call of duty), Michael G Schwern (for testing advice), and the
 573 rest of the CPAN testers (for testing generally).</p>
 574 <p>Special thanks to Tels for suggesting several speed and memory
 575 optimizations.</p>
 576 <p>Additional thanks to:
 577 Edward Avis /
 578 Mattia Barbon /
 579 Tom Christiansen /
 580 Gerrit Haase /
 581 Gurusamy Sarathy /
 582 Jarkko Hietaniemi (again) /
 583 Nikola Knezevic /
 584 John Kominetz /
 585 Nick Ing-Simmons /
 586 Tassilo von Parseval /
 587 H. Dieter Pearcey /
 588 Slaven Rezic /
 589 Eric Roode /
 590 Peter Scott /
 591 Peter Somu /
 592 Autrijus Tang (again) /
 593 Tels (again) /
 594 Juerd Waalboer</p>
 595 <p>
 596 </p>
 597 <hr />
 598 <h1><a name="todo">TODO</a></h1>
 599 <p>More tests.  (Stuff I didn't think of yet.)</p>
 600 <p>Paragraph mode?</p>
 601 <p>Fixed-length mode.  Leave-blanks mode.</p>
 602 <p>Maybe an autolocking mode?</p>
 603 <p>For many common uses of the module, the read cache is a liability.
 604 For example, a program that inserts a single record, or that scans the
 605 file once, will have a cache hit rate of zero.  This suggests a major
 606 optimization: The cache should be initially disabled.  Here's a hybrid
 607 approach: Initially, the cache is disabled, but the cache code
 608 maintains statistics about how high the hit rate would be *if* it were
 609 enabled.  When it sees the hit rate get high enough, it enables
 610 itself.  The STAT comments in this code are the beginning of an
 611 implementation of this.</p>
 612 <p>Record locking with fcntl()?  Then the module might support an undo
 613 log and get real transactions.  What a tour de force that would be.</p>
 614 <p>Keeping track of the highest cached record. This would allow reads-in-a-row
 615 to skip the cache lookup faster (if reading from 1..N with empty cache at
 616 start, the last cached value will be always N-1).
 617
 618 </p>
 619 <p>More tests.
 620
 621 </p>
 622 <table border="0" width="100%" cellspacing="0" cellpadding="3">
 623 <tr><td class="block" style="background-color: #cccccc" valign="middle">
 624 <big><strong><span class="block">&nbsp;Tie::File - Access the lines of a disk file via a Perl array</span></strong></big>
 625 </td></tr>
 626 </table>
 627
 628 </body>
 629
 630 </html>