Menu

#33 current gzip/inflate code in pavuk does not allow huge files

CVS
open-later
Crawler (17)
5
2008-10-30
2008-08-19
Ger Hobbelt
No

Can't download a transfer-compressed HD DVD ISO, because the current gzip/inflate code will load the complete compressed AND decompressed file in memory (in-mem decompressing!).

This should and CAN be fixed for non-parsable files (== anything that's not HTML, CSS or JavaScript).

Discussion

  • Ger Hobbelt

    Ger Hobbelt - 2008-08-19
    • assigned_to: stoecker --> i_a
     
  • Ger Hobbelt

    Ger Hobbelt - 2008-10-23

    pavuk internals do not cope well with overlarge files which do not fit in memory: the code will keep the complete file content in memory for parsing/filtering tasks as well.

    Hence, there are two ways about this: (1) either give pavuk a way to 'stream' overlarge files to disc instead of keeping them in memory, OR (2) deciding this is not a bug but a feature == list this behaviour as 'known and by design'.

    I'm two ways about it. Might have big impact in certain locations and how many people mirror complete DVD images/distros + web pages?

     
  • Ger Hobbelt

    Ger Hobbelt - 2008-10-30
    • status: open --> open-later
     

Log in to post a comment.

Monday.com Logo