Mirror of forum.nim-lang.org

4389 :: need some advice on how to do this

• tsuki (orginal) [2018-11-19T02:08:34+01:00] view original

  Hi everyone! I'm new to nim and have a question about this

  So I have this server on my network that has an encoded video that I download whenever it changes to my device. i used to have a python script that does this, but decided to do it in a new language just for fun. so far I really like nim but I have a problem where if I keep the program running the mem doesnt go down like it does in python so it just goes very high

  this is my code

  
  var  client = newHttpClient()
  
  while true:
      client.downloadfile("http://192.168.1.15/video.mp4", "updated.mp4")
      sleep(60000)
  
  Run

  someone told me I should add client.close but that doesn't fix it and my ram used is still high do I need to use the gcfullcollect function? I tried it and it didn't work... I'm not used to system languages :/

  thanks everyone!

---

• yglukhov (orginal) [2018-11-19T08:36:12+01:00] view original

  Sounds like a bug, but... How high is high? What is the size of the file? What's your os? Have you tested with nim devel version?

---

• Araq (orginal) [2018-11-19T18:37:07+01:00] view original

  Please give us a program that we can reproduce the problem with.

---

• Libman (orginal) [2018-11-21T05:19:20+01:00] view original

  I couldn't reproduce the problem, neither on nim v0.19.0 release nor today's #devel, running under Ubuntu 18-10.

  For my test I created two basic download loop daemons:

  dlHttpClient.nim using Nim stdlib as above -

  import os, times, httpClient
  
  var client = newHttpClient()
  let outName = getAppFilename() & ".out"
  const grabUrl = "http://speedtest.ftp.otenet.gr/files/test1Mb.db"
  
  while true:
    echo "Fetching " & grabUrl & " ..."
    var startTime = cpuTime()
    client.downloadFile grabUrl, outName
    var workSec = (cpuTime() - startTime) * 1000
    echo "... finished in " & $workSec & " sec\n"
    sleep 10*1000
  Run

  dlCurl.nim using libcurl -

  import os, times, libCurl
  
  var outFile: File
  let outName = getAppFilename() & ".out"
  const grabUrl = "http://speedtest.ftp.otenet.gr/files/test1Mb.db"
  
  proc checkCurl(code: Code) =
    if code != E_OK:
      raise newException(AssertionError, "CURL failed: " & $easy_strerror(code))
  
  while true:
    echo "Fetching " & grabUrl & " ..."
    var startTime = cpuTime()
    var curl = libCurl.easy_init()
    discard outFile.open(outName, fmWrite)
    checkCurl curl.easySetOpt(OPT_URL, grabUrl)
    checkCurl curl.easySetOpt(OPT_FOLLOWLOCATION, 1)
    checkCurl curl.easySetOpt(OPT_WRITEDATA, outFile)
    checkCurl curl.easyPerform()
    curl.easyCleanup()
    outFile.close()
    var workSec = (cpuTime() - startTime) * 1000
    echo "... finished in " & $workSec & " sec\n"
    sleep 10*1000
  Run

  After looping for a while, both used a reasonable and stable amount of memory, with the curl version using a little bit more. Pasting from top:

  • dlHttpClient (virt=6220, res=4104, shr=2476)
  • dlCurl (virt=137708 res=9624 shr=8256)

  Replacing "1Mb" with "1Gb" (etc) in grabUrl, memory usage remains reasonable.

---