Rebol：内存不足的问题

Question

我在下面运行了这个程序，其中包含数百个符号的符号列表，有时它说内存不足，而我确实重用了相同的变量，为什么？

base-url: http://www.google.com/finance/historical
download-directory: "askpoweruser/stock-download/google/files/"
column-header: "Time;Open;High;Low;Close;Volume"
#debug: true

symbol_list: parse/all {GOOG AAPL MSFT INDEXDJX:.DJI} " "
ans: ask {symbols by default "GOOG AAPL MSFT INDEXDJX:.DJI": }
if (ans <> "") [symbol_list: parse/all ans " "]
;do code-block/2

foreach symbol symbol_list [
  url0: rejoin [base-url "?q=" symbol]
  dir: make-dir/deep to-rebol-file download-directory

  either none? filename: find symbol ":" [
    filename: symbol
    url: rejoin [url0 "&output=csv"]

    either not error? try [content: read url][
      out-string: copy rejoin [column-header newline]
      quotes: parse/all content ",^/"
      reversed-quotes: reverse quotes

      foreach [v c l h o d] reversed-quotes [
          either not (error? try [d: to-date d]) [
              d: rejoin [d/year "-" d/month "-" d/day]
              append out-string rejoin [d ";" o ";" h ";" l ";" c ";" v newline]
          ][
              ;print [d "is not a date"]
              ;input
          ]

      ]
      filename: rejoin [filename "_" "1440"]
      write to-rebol-file rejoin [dir filename ".csv"] out-string
      print filename
    ][
      print ["Error for symbol" symbol]
    ]
  ][
    filename: replace/all replace/all filename ":" "" "." ""
    out: copy []
    for i 0 1 1 [
      p: i
      url: rejoin [url0 "&start=" (p * 200) "&num=" ((p + 1) * 200)]
      content: read url
      rule: [to "<table" thru "<table" to ">" thru ">"
          to "<table" thru "<table" to ">" thru ">"
          to "<table" thru "<table" to ">" thru ">"
          copy quotes to </table> to end
      ]
      parse content rule

      parse quotes [
          some [to "<td" thru "<td" to ">" thru ">" [copy x to "<" | copy x to end] (append out replace/all x "^/" "")]
          to end
      ]
      if #debug [
          write/lines to-rebol-file rejoin [dir filename "_" p ".html"] quotes
      ]

    ]
    if #debug [
        write to-rebol-file rejoin [dir filename "_temp" ".txt"] mold out
    ]
    out-string: copy rejoin [column-header newline]
    out: reverse out    

    foreach [v c l h o d] out [
       d: parse/all d " ,"
       d: to-date rejoin [d/4 "-" d/1 "-" d/2]
       d: rejoin [d/year "-" d/month "-" d/day]
       append out-string replace/all rejoin [d ";" o ";" h ";" l ";" c ";" v newline] "," ""
    ]
    filename: rejoin [filename "_" "1440"]
    write to-rebol-file rejoin [dir filename ".csv"] out-string
    print filename
  ]

]

要获取符号列表，您可以使用这个（rebol 在字母 H 之前崩溃了）：

alphabet: [A B C D E F G H I J K L M N O P Q R S T U V W X Y Z]
symbol-list: copy []
rule: [
    to <table class="quotes">
    some [ to {<A href="/stockquote} to ">" thru ">" copy symbol to "<" (append symbol-list symbol)]
    to </table>
]
foreach letter alphabet [
    content: read to-url rejoin ["http://www.eoddata.com/stocklist/NYSE/" letter ".htm"]
parse content rule
probe symbol-list
write/append %askpoweruser/stock-download/symbol-list-nyse.txt mold symbol-list
]

Answer 1

您可以在一个循环中放置一个“STAT 函数”，以尝试找出是否以及在何处发生内存泄漏。

内存不足错误通常发生在以下情况之一或类似的情况下：

• 附加到在带有循环的函数开始时未清除或复制的系列

• 在树和整个树外部引用某些（单个？）子元素的情况下，完整的数据树不会重置为无（在每个叶子和分支）数据块最终被困在内存中，无法自行释放

• VID 面包含对完整样式表的引用，因此打印窗口大型应用程序的窗口通常会失败。）。

• 某些堆栈溢出（无休止的递归或循环）有时会被错误地报告为内存错误。

• 单个项目的分配呈指数增长......就像图像一样！在两个轴上将每次传递乘以 10 时的分配...有效地增加了两个数量级，这通常在 n*10k 范围内的数字上失败。

• GC 中最大的项目有时永远不会根据次优 R2 GC 取消分配（大图像可能有此症状）。

• 递归解析规则正在创建数据，并且单个规则是无限的（在像 [rule | none ] 这样的规则上发生得非常快。在这种情况下，none 有效地相当于永远。

Answer 2

它工作没有问题。但一些注意事项：

instead:
read to-url rejoin ["http://www.eoddata.com/stocklist/NYSE/" letter]
you can use just:
read join http://www.eoddata.com/stocklist/NYSE/ letter

instead:
either not cond [1][2]
you should use:
either cond [2][1]

instead:
download-directory: "askpoweruser/stock-download/google/files/"
dir: make-dir/deep to-rebol-file download-directory
write to-rebol-file rejoin [dir filename "_temp" ".txt"] mold out

use:
download-directory: %askpoweruser/stock-download/google/files/
dir: make-dir/deep download-directory
write rejoin [dir filename %_temp.txt] mold out

what the hell means?!:
for i 0 1 1 [
      p: i
      url: rejoin [url0 "&start=" (p * 200) "&num=" ((p + 1) * 200)]
..