[Bioperl-l] Bio::DB::GenBank and proxy

[simon andrews (BI)]

> -----Original Message-----
> From: William Kenworthy [mailto:billk at iinet.net.au] 
> Sent: 12 August 2003 12:51
> To: simon andrews (BI)
> Cc: BioPerl List
> Subject: RE: [Bioperl-l] Bio::DB::GenBank and proxy
> 
> 
> Unfortunately, this didn't work for me:
> 
> declare -x http_proxy="http://localhost:8081"
> wdk at rattus tmp $ ./t.pl
> Attempt to bless into a reference at 
> /usr/lib/perl5/site_perl/5.8.0/LWP/UserAgent.pm line 221. 
> wdk at rattus tmp $ 

I take it this is after you'd made the changes I put in my bug report?  I can't really diagnose this as I think I have a different version of UserAgent.pm to you.  Line 221 in mine sets up a cookie jar, which doesn't sound right.

When you changed the two bioperl modules were the lines you altered already calls to the new method of LWP::UserAgent?  The line numbers reported may only be correct for the 1.2 release of bioperl.

Can you run this small prog and see what you get (please unwrap the long line on the get call).

#!/usr/bin/perl -w
use strict;
use LWP::UserAgent;

my $ua = LWP::UserAgent -> new (env_proxy => 1,
				keep_alive => 1,
				timeout => 30);

my $response = $ua -> get('http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?retmode=text&rettype=gb&db=nucleotide&tool=bioperl&id=gb%7CAL022723&usehistory=n');

my $text = $response -> as_string;

print substr($text,0,1000) , "\n";
#################################################################

You should hopefully get something like this:

#####################
HTTP/1.0 200 OK
Date: Tue, 12 Aug 2003 13:16:55 GMT
Via: 1.1 eutils.ncbi.nih.gov
Server: Apache
Content-Type: text/plain
Client-Date: Tue, 12 Aug 2003 19:17:13 GMT
Client-Response-Num: 1
Proxy-Connection: close
X-Cache: MISS from eutils.ncbi.nih.gov
X-Cache: MISS from BBSRC-wwwcache-service

LOCUS       HS377H14              148834 bp    DNA     linear   PRI 14-SEP-2001
DEFINITION  Human DNA sequence from clone RP3-377H14 on chromosome
            6p21.32-22.1. Contains the HLA-G gene for major [andrewss at bilin2 Test]$ ./uatest.pl
HTTP/1.0 200 OK
Date: Tue, 12 Aug 2003 13:17:13 GMT
Via: 1.1 eutils.ncbi.nih.gov
Server: Apache
Content-Type: text/plain
Client-Date: Tue, 12 Aug 2003 19:17:32 GMT
Client-Response-Num: 1
Proxy-Connection: close
X-Cache: MISS from eutils.ncbi.nih.gov
X-Cache: MISS from BBSRC-wwwcache-service

LOCUS       HS377H14              148834 bp    DNA     linear   PRI 14-SEP-2001
DEFINITION  Human DNA sequence from clone RP3-377H14 on chromosome
            6p21.32-22.1. Contains the HLA-G gene for major histocompatibility
            complex class I G (HLA 6.0) an MHC class I pseudogene, an RPL7A
            (60S Ribosomal Protein L7A) pseudogene, a gene for a novel MHC
            class 1 protein, an interferon-inducible protein 1-8U pseudogene,
            an RPL23A (60S Ribosomal Protein L23A) pseudogene, an HCGIX
            pseudogene, an MICB or PERB11.1 pseudogene,the HLA-F gene for major
            histocompatibility complex class I F (CDA12), and four P5-1
            pseudogenes. Con
#####################

If this works can you go back and check the amendments you made to the two bioperl scripts as this is all they are doing.  If it still fails then its not a bioperl problem as such, but we can still try to track it down.

 
> Noticed another thing: the cache (squid in my case) always 
> reports a "miss", and fetches direct.

That's probably right.  Since you're fetching potentially dynamic content (albeit through a GET request) you may find that squid refuses to cache it and will refetch it every time.  Have a look in the squid documentation and there may be some way to tell it that it should cache dynamic GET requests as well.

Hope this helps

Simon.