Re^2: A case where Mechanize works, LWP doesn't
by Anonymous Monk on Feb 09, 2010 at 20:19 UTC
|
Sigh. I figured I made a typo or forgot a call. Yep loaded wireshark. I guess LWP isn't as straight forward as I thought. Is LWP normally this difficult or is this just a bad one-off. I cannot imagine debugging a large program via sniffer traces. | [reply] |
|
|
WWW::Mechanize uses LWP, so your question makes no sense.
| [reply] |
|
|
Read the original post which shows the Mechanize program works while the LWP program doesn't.
| [reply] |
|
|
|
|
|
|
|
|
|
Re^2: A case where Mechanize works, LWP doesn't
by dneedles (Sexton) on Feb 10, 2010 at 16:12 UTC
|
Ok. Interesting results on the Sniffer trace. It appears Mechanize is handling a redirect while LWP is not. Also the encapsulation of the POST on Mechanize is application/x-www-form-urlencoded while LWP just has variable: value in separate lines. So it appears Mechanize is doing much more work under the covers. I'll see how I can manipulate LWP to mimic Mechanize. | [reply] |
|
|
| [reply] |
|
|
Great questions. The reason is two fold.
1. Limit change. The existing PERL installation doesn't use Mechanize and I would rather not introduce it primarily to keep the politics under control.
2. POE integration. I have seen much on LWP and POE but nothing on Mechanize and POE. I'd rather go with a more proven route.
Correct me if I am wrong (with a URL, ISDN or other tangible reference) especially on #2
| [reply] |
|
|
|
|
|
|
|
|
Actually found out this isn't the case. Mechanize has blocking calls. POE is more a cooperative scheduler along the lines of what Windows 3.1 was. So POE won't work well with Mechanize like it will with LWP. So I am back to using the sniffer to figure out how to code LWP with enough of Mechanize so the pages will return.
| [reply] |