[Open-bio-l] Toolkits and the new eutils policies

[Chris Fields]

I'll cc this for the others in BioPerl, Biopython, Bioruby, Ensembl, etc so they know and can pass it around.  Thanks Eric, happy to have that clarified.

chris

On Mar 25, 2010, at 1:06 PM, sayers wrote:

> Hi Chris,
> 
> Thanks for your note. Organizations such as yours were in our minds when we considered this policy change, and perhaps we need to consider improving the wording of the policy so that it is clearer on the points that you raise. 
> 
> Our requests for registered values for &tool and &email are directed at "end-developers" of software, not necessarily at toolkit developers such as yourself, whose products may be used by end-developers. We appreciate any efforts you can make to encourage and facilitate the use of &tool and &email by end-developers, but the act of registering values of these parameters is the responsibility of the end-developers. The value of &email should be a contact address for the end-developer, and ideally the value of &tool should be the the name of the software package that uses BioPerl or other toolkits. 
> 
> My suggestion would be similar to what you mention in your second paragraph: to leave both &tool and &email without values and give a warning or some equivalent to developers if they fail to set them.
> 
> Regarding blocks, we block on the basis of IP addresses only. The essence of the new policy is that we are far more likely to block requests that do not have registered values of &tool and &email. If we need to block activity that does have a registered &tool or &email value, we will only block those IPs that are causing the abusive activity, not all IPs using that &tool/&email value.
> 
> Thanks again for your comments and please let me know if you have further questions.
> 
> Regards,
> Eric
> ___________________
> Eric W. Sayers, PhD
> NCBI/NLM/NIH
> 45 Center Drive, MSC 6511
> Bldg 45, Room 4AN.44C
> Bethesda, MD 20892
> sayers at ncbi.nlm.nih.gov
> 
> 
> 
> 
> On Mar 24, 2010, at 11:31 AM, Chris Fields wrote:
> 
>> To whom it may concern,
>> 
>> Just want to get some clarification from the eutils folks with the changes.  I'm a core developer for the BioPerl toolkit and and in collaboration with several other Bio* toolkits (BioPython, BioRuby, BioJava).  We all have interfaces to eutils, either via the standard interface, SOAP-based, or both.
>> 
>> We're seeking a clarification regarding the new rules, specifically the rules concerning 'tool' and 'email'.  Per the rules proposed in Dec. 2009, many of us have already implemented changes to address them.  As we're basically language-specific toolkits (classes, modules, etc) that are used for developing other applications, we have taken the stance that the end-user will need to start providing at least the email in order to properly use the eutils-specific tools, with a warning issued otherwise.  This is based on several reasons, foremost being the toolkits are very widely used (so may be spread over potentially thousands of IPs) and are being used in a large number of downstream applications.  As an example of the latter, BioJava is used in several free and commercial applications, such as Taverna and Geneious.
>> 
>> Currently, with BioPerl and Biopython, we set the 'tool' parameter specifically to the toolkit name by default.  This parameter can be overridden.  However, from reading the newest rules it appears that each tool (and thus each toolkit) should have one common email, no more.  This also appears to make the assertion that users using these toolkits (with 'tool' set and using the relevant emails) may essentially be tied down if one IP decides to abuse the rules.
>> 
>> This unfortunately makes the incorrect generalization that each tool is created by one developer, and simply does not make sense in large collaborative projects such as ours, where the software is used primarily in downstream applications and scripts.  Setting the tool could be beneficial to the development team, but at the same time it could be a tremendous hindrance.
>> 
>> So, what exactly should we do?  Do we set the 'tool' by default, or leave it to the user?  Similarly, how do we treat 'email'?  If we do set them, would the entire group of users for that tool be blocked if one end-user abuses the system?  Should we leave it up to the user to register themselves (both tool and email)?  So, we're at an impasse and really need your help.
>> 
>> Sincerely,
>> 
>> chris
>> 
>> 
>> Christopher Fields
>> Core Developer, BioPerl Project
>> IGB Postdoctoral Fellow
>> Genomics of Neural & Behavioral Plasticity
>> University of Illinois Urbana-Champaign
>> Institute for Genomic Biology
>> 1206 W. Gregory Dr. , MC-195
>> Urbana, IL 61801
>> 
>