Difference between revisions of "BibTestLM"

From Planck Legacy Archive Wiki
Jump to: navigation, search
 
(38 intermediate revisions by 2 users not shown)
Line 1: Line 1:
+
Bellow you will find a summary of the current status of referencing in the ES.
  
 +
* The solution proposed obeys the following requirements:
 +
 +
#It needs to allow the use of bibtex. Ideally the use of the Planck publications bibtex file with no pre processing whatosever, or in case needed a little bit of clen up.
 +
#Needs to allow the storage of the full bibliography in a form that is ES wide, i.e. all the references in the bibtex file should be usable from any page in the wiki not only in the page where it was uploaded, otherwise the contributors will be forced to input the bibtex records for every page where they are required.
 +
#The use of the citation mechanism, should be simple and if possible using the same key for each bibtex record as the one use for the papers;
 +
#Once a bibtex record is cited in a page it should appear somewhere in the page;
 +
#A link to an online version of the paper should be part of the paper details appearing as the result of the citation.
 +
 +
* More requirements can be added such as the ones implied by Ken and Alain (I will discuss those later).
 +
 +
It will probably help to split the following in two cases non-Planck papers and Planck collaboration papers. The main difference being from the practical point of view the fact that Planck papers are universally available while non-Planck papers are only available to subscribers of the journals where they were published.
 +
#non-Planck Papers can be cited with the ''cite'' tag as in <cite>#ferreira2000</cite>. Clicking on the number you will be taken to the bottom where you can see the full reference. Clicking on the title on the full reference you will be taken to an url provided in the bibtex record. Even if you have access to the journal it is not possible to open a pdf in a given page.
 +
#For Planck papers you can also use the same mechanism as above but you can do more. Using the templates I provided some time ago you can actually open the pdf on a given page. I still think it is important to have the full reference accessible to the user and you could possible use these two mechanisms together like so in this paper <cite>#planck2011-5-2a</cite> {{PEarly|10|page=4|link=we discuss something interesting}}
 +
 +
This is what is available at the moment from your comments it looks like it does not satisfy your requirements. Unfortunately until the release I find it very difficult (and I really mean impossible) to have something a lot more sophisticated. Personally I recognize this is not perfect but I don't think it is a complete disaster and therefore unacceptable.
 +
 +
I reply to your comments below.
  
{{PPreLaunch|10||link=and we get a link here <cite>#dupac2005,#hanson2009</cite>}} but it looks a bit odd.
 
  
 
Laurent own test:
 
Laurent own test:
Line 8: Line 24:
  
  
LM replies:
 
Please disregard the example above... I wrote it while I was playing with different possibilities. Instead look at this. I can cite a paper like this <cite>#planck2011-1-1</cite>. To add references to a page you include the lines at the end of said page (the biblio stuff). Now by clicking on the number inside square brackets you will be taken to the reference at the end of the page. In this case nothing happens because the page is to small for scrolling to be necessary. If you now clik on the name of the paper you will be taken to the corresponding url.
 
 
Note that there is '''no''' need to enter the details of each bib record everytime you need to cite one. All the records are stored int he References page.
 
  
 
AC: we are using a MediaWiki engine. I saw with Laurent that on current WikiPedia pages that we can have a window containing informations (suggestion: bib entry)
 
AC: we are using a MediaWiki engine. I saw with Laurent that on current WikiPedia pages that we can have a window containing informations (suggestion: bib entry)
 
and link (e.g. to the PDF)
 
and link (e.g. to the PDF)
  
Ken's two Euro-cents: I would like to have some indication of to what we are referring to in the text itself -- often I'll recognize it, and can ignore it. It should be short, but not too short, if you know what I mean. A simple number is a bit too short -- I can't recognize it. I really dislike having to go somewhere else and then decide what to do. It's a distraction. I want to be able to decide right while I'm reading the text if I should click and go somewhere else, or stay where I am. So, I think this means that I like Laurent's Own Test above, but instead of just seeing a number, I should see something like ([[http://inspirehep.net/record/295501|Vibert et al. 1990]]). Then, if I click the reference, or if necessary, the little icon next to it, I will be taken to the astro-ph/ApJ/A&A abstract page...
+
<span style="color:Red">LM:</span> Yes you are right. But:
 +
*We are using a rather old version of Medaiwiki and the pop ups with the details do no seem to work. Why don't we upgrade? this has been pending but it is not our decision. If we want LDAP access control we have to go with what our computer services here have on offer. If our computer services cannot guarantee security and stability I don't think it is a good idea to upgrade.
 +
*You will notice that the references are not using bibtex. Should we drop the bibtex requirement? Or maybe we can convert bibtex to use with the same [http://en.wikipedia.org/wiki/Template:Citation tempalte] wikipedia is using? Fine, but then see next
 +
*The bibliographic references in wikipedia are entered on a per page basis. This means there is no central repository of references which then have to be entered on envery page they are used.
 +
 
 +
Ken's two Euro-cents: I would like to have some indication of to what we are referring to in the text itself -- often I'll recognize it, and can ignore it. It should be short, but not too short, if you know what I mean. A simple number is a bit too short -- I can't recognize it. I really dislike having to go somewhere else and then decide what to do. It's a distraction. I want to be able to decide right while I'm reading the text if I should click and go somewhere else, or stay where I am. So, I think this means that I like Laurent's Own Test above, but instead of just seeing a number, I should see something like ([[http://inspirehep.net/record/295501 Vibert et al. 1990]]). Then, if I click the reference, or if necessary, the little icon next to it, I will be taken to the astro-ph/ApJ/A&A abstract page...
 +
 
 +
<span style="color:Red">LM:</span> This would be nice to have but may not be as good as you suggest all the time as for Planck most references would then be Ade et al.
 +
 
 +
'''LM'''s conclusion for now: We have many requirements and they all make sense but at the moment we cannot implement them all. I think what we have available is not so bad that it cannot be used. There is a lot of scope for improvement but it is not unusable.
  
 
<biblio force=false>
 
<biblio force=false>
 
#[[References]]
 
#[[References]]
 
</biblio>
 
</biblio>

Latest revision as of 14:08, 8 February 2013

Bellow you will find a summary of the current status of referencing in the ES.

  • The solution proposed obeys the following requirements:
  1. It needs to allow the use of bibtex. Ideally the use of the Planck publications bibtex file with no pre processing whatosever, or in case needed a little bit of clen up.
  2. Needs to allow the storage of the full bibliography in a form that is ES wide, i.e. all the references in the bibtex file should be usable from any page in the wiki not only in the page where it was uploaded, otherwise the contributors will be forced to input the bibtex records for every page where they are required.
  3. The use of the citation mechanism, should be simple and if possible using the same key for each bibtex record as the one use for the papers;
  4. Once a bibtex record is cited in a page it should appear somewhere in the page;
  5. A link to an online version of the paper should be part of the paper details appearing as the result of the citation.
  • More requirements can be added such as the ones implied by Ken and Alain (I will discuss those later).

It will probably help to split the following in two cases non-Planck papers and Planck collaboration papers. The main difference being from the practical point of view the fact that Planck papers are universally available while non-Planck papers are only available to subscribers of the journals where they were published.

  1. non-Planck Papers can be cited with the cite tag as in #ferreira2000. Clicking on the number you will be taken to the bottom where you can see the full reference. Clicking on the title on the full reference you will be taken to an url provided in the bibtex record. Even if you have access to the journal it is not possible to open a pdf in a given page.
  2. For Planck papers you can also use the same mechanism as above but you can do more. Using the templates I provided some time ago you can actually open the pdf on a given page. I still think it is important to have the full reference accessible to the user and you could possible use these two mechanisms together like so in this paper #planck2011-5-2a we discuss something interesting

This is what is available at the moment from your comments it looks like it does not satisfy your requirements. Unfortunately until the release I find it very difficult (and I really mean impossible) to have something a lot more sophisticated. Personally I recognize this is not perfect but I don't think it is a complete disaster and therefore unacceptable.

I reply to your comments below.


Laurent own test:

  • This example here #dupac2005,#hanson2009 doesn't gives the titles of the papers in the text itself. And this forces to have at the bottom of each wiki page all papers referecned in the page.


AC: we are using a MediaWiki engine. I saw with Laurent that on current WikiPedia pages that we can have a window containing informations (suggestion: bib entry) and link (e.g. to the PDF)

LM: Yes you are right. But:

  • We are using a rather old version of Medaiwiki and the pop ups with the details do no seem to work. Why don't we upgrade? this has been pending but it is not our decision. If we want LDAP access control we have to go with what our computer services here have on offer. If our computer services cannot guarantee security and stability I don't think it is a good idea to upgrade.
  • You will notice that the references are not using bibtex. Should we drop the bibtex requirement? Or maybe we can convert bibtex to use with the same tempalte wikipedia is using? Fine, but then see next
  • The bibliographic references in wikipedia are entered on a per page basis. This means there is no central repository of references which then have to be entered on envery page they are used.

Ken's two Euro-cents: I would like to have some indication of to what we are referring to in the text itself -- often I'll recognize it, and can ignore it. It should be short, but not too short, if you know what I mean. A simple number is a bit too short -- I can't recognize it. I really dislike having to go somewhere else and then decide what to do. It's a distraction. I want to be able to decide right while I'm reading the text if I should click and go somewhere else, or stay where I am. So, I think this means that I like Laurent's Own Test above, but instead of just seeing a number, I should see something like ([Vibert et al. 1990]). Then, if I click the reference, or if necessary, the little icon next to it, I will be taken to the astro-ph/ApJ/A&A abstract page...

LM: This would be nice to have but may not be as good as you suggest all the time as for Planck most references would then be Ade et al.

LMs conclusion for now: We have many requirements and they all make sense but at the moment we cannot implement them all. I think what we have available is not so bad that it cannot be used. There is a lot of scope for improvement but it is not unusable.

<biblio force=false>

  1. References

</biblio>

Explanatory Supplement