Optimize Tutorial

---

Since the new architecture support multi-connection I test on my home, it can use all resources on my network (the broadband on my home is 100Mb/s):

36MB file can be finished on about 3.082s:

---

1. process.non-separate=true

Through set process.non-separate=true in filedownloader.properties to let the FileDownloader service runs in the main process.

Advantage

Improve download speed.

• Reduce CPU consume for IPC and the separate process maintain.
• Reduce I/O consume for IPC.

Disadvantage

The FileDownloader service runs in the main process.

If your need your download works in the background what means you want the FileDownloader service isolated in the background which lifecycle is decoupling with the main process, in this case, you need to be very careful to consider whether this optimization is necessary.

2. BaseDownloadTask#setCallbackProgressIgnored(void) or FileDownloadQueueSet#ignoreEachTaskInternalProgress(void)

Ignore all callbacks of FileDownloadListener#progress during the entire process of downloading.

Advantage

Improve download performance.

• Reduce CPU and I/O consume for transfer the progress callbacks.

Disadvantage

You will not receive any FileDownloadListener#progress method callback during the entire process of downloading.

If you indeed care about the downloading progress during the downloading process, maybe you need alternative BaseDownloadTask#setCallbackProgressTimes and BaseDownloadTask#setCallbackProgressMinInterval.

3. FileDownloadBufferedOutputStream with file.non-pre-allocation=true

NOT RECOMMEND ANYMORE, because FileDownloadBufferedOutputStream can't support seek, what means FileDownloader can't download file on the multi-connections(the multi-connections feature is supported since v1.5.0).

Through initializes the FileDownloader with:

FileDownloader.init(getApplicationContext(),
  new DownloadMgrInitialParams.InitCustomMaker().
    outputStreamCreator(new FileDownloadBufferedOutputStream.Creator()));

and set file.non-pre-allocation=true in filedownloader.properties to let FileDownloader doesn't pre-allocate the entire length file in the very beginning of download, because FileDownloadBufferedOutputStream can't support the seek function.

Advantage

Improve download speed.

• The I/O rule in the BufferedOutputStream is better than the rule in the default output stream component RandomAccessFile for FileDownloader(More statistic ref here).

Disadvantage

Since the FileDownloadBufferedOutputStream uses BufferedOutputStream as the output stream component which doesn't support the seek function, if FileDownloader pre-allocates file size at the beginning of the download, it will can not be resumed from the breakpoint. If you need FileDownloader can pre-allocates the entire length file in the very beginning of download for preventing occur FileDownloadOutOfSpaceException during the downloading process, you have to give up this optimization to use the default output stream component, or implement you own output stream component.

4. Custmoize your own other component

More detail please move to here.

---

Others

• FileDownloader#disableAvoidDropFrame()
• BaseDownloadTask#addFinishListener(FinishListener) instead of BaseDownloadTask#setListener(FileDownloadListener)
• BaseDownloadTask#setSyncCallback(true)
• download.min-progress-step、download.min-progress-time in filedownloader.properties