But if I understand the situation correctly, for six years Zotero stored the pubmed url automatically, a single user (approved by the lead), removed that feature, and now even a temporary rollback is "not viable". I think that you and I disagree on the meaning of the term "viable".
I do appreciate the other comments in the thread. Hopefully there will be a way to solve this issue soon.
I believe that the _primary_ purpose is to allow users to capture bibliographic data and use it to add reference citations to manuscripts. The problem with including the PubMed URL where you want it to be is that the PubMed URL will be included in any citation instead of the correct URL that allows direct access to the full text item.
Clearly, you are strongly invested in your desire to revert to the old way. However, I don't think that high emotion is the way to convince the developers and volunteers that your way is best.
I don't know how public the comments have been complaining about the PubMed database URL being used instead the correct URL for citation. However, using the PubMed URL required manually editing the manuscript to change to the proper citation form.
I think that few would object to including a linkable PubMed URL in a note field or someplace similar. I believe that you have made your point that having this feature makes sharing links to PubMed easy. However, I personally don't understand how sharing a PubMed URL makes it easier for the recipient of the shared record when a "proper" URL that goes to the item on a publisher's site would allow her or him to view the original record. To me, having a link to the original item is more important than having a link to a database record.
I don't know how public the comments have been complaining about the PubMed database URL being used instead the correct URL for citation. However, using the PubMed URL required manually editing the manuscript to change to the proper citation form.
that exactly. And it makes absolutely no sense to revert back to a solution producing that type of import for all users.
That said, I do understand that and why some users want the linked PubMed article as part of their citation and I and others have suggested ways that we'll be able to do that in the future - I believe Dan in another thread suggested an additional field as a third option (apart from Simon's above and the PMID version I suggest), so it's not like we're ignoring that demand.
I've also offered to help with individualized workarounds which require in the vicinity of 5 lines of code, though obviously my enthusiasm for helping someone decreases with the amount of invective thrown at me...
If you really really want to get the links back in the URL field, here is a custom version of the PubMed translator that is functional as of this writing. This brings back old behavior. I do not recommend using it and I do not promise to remember to keep it up to date. The translator will not update automatically even if I do update it at the given link, so you will have to come back and update manually if things break.
To install this, save the linked file to your {Zotero Data Directory}/translators directory. I suggest naming it "NCBI PubMed (with PubMed URL).js" but you can chose whatever you like.
While I understand the necessity of this change in order to adapt a "clean" database that corresponds to standards I completely fail to understand why an option to keep the old behavior was not added.
In terms of ease of use this change is all down hill. I frequently check the abstract of an item in Zotero to be sure I'm citing the correct paper, then if needed, I open the corresponding pdf file / go to the publishers page. Note that abstracts are often much easier to read on Pubmed (or other indexing pages) than on the publisher webpage.
Before I could simply double click the item in Zotero anywhere - now I'm left with having to click a small arrow follow by a double click. This has annoyed me for months and the first thing I did when the change went into effect was to check if there was a new option I missed...
I see above a comment about "this seems only to have affected 4 users" and that makes me sad. Those 4 users were the ones that spoke up for everyone else who do not frequent forums. Comments about: This is open source, you can fix it yourself - here, use this (soon to be) outdated translator are also not helpful and seem more arrogant than trying to actually solve the problem at hand: That usability of Zotero went down with this change for a lot of users.
I really really love the work done on Zotero and despite this change it is still the best reference manager out there, its just slightly more annoying to use now.
In Zotero 3.0.12, it will be possible to define translator-specific hidden preference. The hidden pref suggested by Simon in this thread might be added then (?).
(an option wasn't introduced because Zotero doesn't have the code for such an option in place. Much of the productive discussion in this thread is about how to create such an option).
Yes, I'm aware there is work being done on this and that hopefully it will be possible to get the original behavior back. My gripe (although I'm very thankful for Zotero and all the work being put into it! :-) is why it was considered perfectly ok to change the behavior silently instead of making it optional.
Thanks for continuing to work on this, getting the "feature" (or bug ;-) back as an option would be extremely welcome!
I mean - what do you do? If you don't change the behavior it remains wrong for longer - including importing wrong data for most users (I can't emphasize enough that that's really a big deal, which is why we've said no on all the calls to revert this) - if you do change it you annoy some users as this thread demonstrates. I don't really see what could have been done differently.
I don't think we anticipated the intensity of feelings about this, but even with that, I don't really see a happy middle option.
In terms of ease of use this change is all down hill. I frequently check the abstract of an item in Zotero to be sure I'm citing the correct paper, then if needed, I open the corresponding pdf file / go to the publishers page. Note that abstracts are often much easier to read on Pubmed (or other indexing pages) than on the publisher webpage.
I take it you do not find it convenient to read the abstract in Zotero itself if you find the need to go to PubMed to read it. I think that's an issue we would be interested in addressing. I think I find the abstract a bit crowded and not as convenient to get to as it could be. IMO if we introduced a frame on the bottom of the list that would immediately show the abstract for a selected item, it would be better. We can also use that frame to show a preview of a PDF at some point in the future (I think Simon wants to do this to some extent anyway)
@aurimas I certainly agree that viewing the abstract in Zotero's right column is difficult. If there were some enhancement of the user interface that improved the abstract view I would really be pleased. I don't know about the frame idea. It seems that it could get in the way. Perhaps an option could be added to the right-click menu on the record in the middle column could bring up a formatted read-only version of the citation and abstract.
I imagine that the frame would be collapsible, just like the tag panel. I typically use Zotero Standalone, so screen real estate is not an issue, but I can see how this could be a bit much in Firefox. Maybe a keyboard shortcut could collapse/minimize this frame.
Yes, I meant the FF version. The SA version would be less a problem. This will be a valuable improvement. I trust that the UI display issues can be worked out satisfactorily. Not only is the current abstract view rather crowded, it is "live". I have scrolled and moused over the abstract to view more of it and deleted part of it by accident.
What about the double-click behavior? That should be a very quick solution to some the issue hartvig complains about most. Two separate issues: 1. If an item doesn't have a DOI or URL and it's only attachment is a link, currently double-clicking does nothing. It should open the link. This is IMHO a no-brainer. 2. On the order in which items are opened, I'd suggest attached links should take preference over resolved DOIs. This may be more controversial.
It seems that most of you don't use LaTeX/BibTeX, but FWIW, this change results in no "url" field in bibtex entries for pubmed abstracts, hence no url is generated by bibtex in the references list.
Please consider that this breaks the previous functionality and also uniformity, compared with pulling from almost every other source actually storing a url.
I don't mean to be rude, as I'm grateful for the functionality of Zotero, but I don't think the repercussions of this change were thoroughly considered before implementation. Why didn't you simply allow a preferences option for the functionality to change for those who wanted it the new way, and an option for it to remain the same for the rest of us?
FWIW, the PMID is now mapped to CSL, so it would be possible to write a citation style that provides a URL. Let me know if that's still of interested to people and which style would be good as a basis.
I'm not entering the rest of the debate again, everything on this has been said.
Thanks for being responsive to all the different views out there.
My workflow is Zotero --> LyZ --> LyX --> HTML/PDF, which works surprisingly well, save that LyZ currently doesn't figure out the location of the URL as an attachment to pubmed entries. This is obviously not Zotero's responsibility, as LyZ is coded to be an add on to Zotero, not vice versa.
It's probly possible to recode LyZ a bit to make it work with this change, since in principle the information is there. It's beyond my expertise at the moment, however.
Upgrade Storage
http://www.zotero.org/support/preferences/hidden_preferences
It may not be productive to continue our dialog.
But if I understand the situation correctly, for six years Zotero stored the pubmed url automatically, a single user (approved by the lead), removed that feature, and now even a temporary rollback is "not viable". I think that you and I disagree on the meaning of the term "viable".
I do appreciate the other comments in the thread. Hopefully there will be a way to solve this issue soon.
Clearly, you are strongly invested in your desire to revert to the old way. However, I don't think that high emotion is the way to convince the developers and volunteers that your way is best.
I don't know how public the comments have been complaining about the PubMed database URL being used instead the correct URL for citation. However, using the PubMed URL required manually editing the manuscript to change to the proper citation form.
I think that few would object to including a linkable PubMed URL in a note field or someplace similar. I believe that you have made your point that having this feature makes sharing links to PubMed easy. However, I personally don't understand how sharing a PubMed URL makes it easier for the recipient of the shared record when a "proper" URL that goes to the item on a publisher's site would allow her or him to view the original record. To me, having a link to the original item is more important than having a link to a database record.
That said, I do understand that and why some users want the linked PubMed article as part of their citation and I and others have suggested ways that we'll be able to do that in the future - I believe Dan in another thread suggested an additional field as a third option (apart from Simon's above and the PMID version I suggest), so it's not like we're ignoring that demand.
I've also offered to help with individualized workarounds which require in the vicinity of 5 lines of code, though obviously my enthusiasm for helping someone decreases with the amount of invective thrown at me...
To install this, save the linked file to your {Zotero Data Directory}/translators directory. I suggest naming it "NCBI PubMed (with PubMed URL).js" but you can chose whatever you like.
In terms of ease of use this change is all down hill. I frequently check the abstract of an item in Zotero to be sure I'm citing the correct paper, then if needed, I open the corresponding pdf file / go to the publishers page. Note that abstracts are often much easier to read on Pubmed (or other indexing pages) than on the publisher webpage.
Before I could simply double click the item in Zotero anywhere - now I'm left with having to click a small arrow follow by a double click. This has annoyed me for months and the first thing I did when the change went into effect was to check if there was a new option I missed...
I see above a comment about "this seems only to have affected 4 users" and that makes me sad. Those 4 users were the ones that spoke up for everyone else who do not frequent forums. Comments about: This is open source, you can fix it yourself - here, use this (soon to be) outdated translator are also not helpful and seem more arrogant than trying to actually solve the problem at hand: That usability of Zotero went down with this change for a lot of users.
I really really love the work done on Zotero and despite this change it is still the best reference manager out there, its just slightly more annoying to use now.
The hidden pref suggested by Simon in this thread might be added then (?).
This won't be possible to implement in the short-term, though, because of current limitations in how syncing works.
Thanks for continuing to work on this, getting the "feature" (or bug ;-) back as an option would be extremely welcome!
I don't think we anticipated the intensity of feelings about this, but even with that, I don't really see a happy middle option.
Two separate issues:
1. If an item doesn't have a DOI or URL and it's only attachment is a link, currently double-clicking does nothing. It should open the link. This is IMHO a no-brainer.
2. On the order in which items are opened, I'd suggest attached links should take preference over resolved DOIs. This may be more controversial.
Please consider that this breaks the previous functionality and also uniformity, compared with pulling from almost every other source actually storing a url.
I don't mean to be rude, as I'm grateful for the functionality of Zotero, but I don't think the repercussions of this change were thoroughly considered before implementation. Why didn't you simply allow a preferences option for the functionality to change for those who wanted it the new way, and an option for it to remain the same for the rest of us?
I'm not entering the rest of the debate again, everything on this has been said.
My workflow is Zotero --> LyZ --> LyX --> HTML/PDF, which works surprisingly well, save that LyZ currently doesn't figure out the location of the URL as an attachment to pubmed entries. This is obviously not Zotero's responsibility, as LyZ is coded to be an add on to Zotero, not vice versa.
It's probly possible to recode LyZ a bit to make it work with this change, since in principle the information is there. It's beyond my expertise at the moment, however.