Automatically adding "online"

[wg030]

When creating the PR #115 I startet to wonder if it makes sense to automatically add "online" when printing an entry which has got an urldate field. Would it not make more sense to let the user decide if he watns that to be printed or not?

For example, the example that I added from the latest version of the norm is not printed with "online" in the text of the norm, but with the current state of this package there is now way to suppress and reproduce that.

I also wondered if howpublished is the correct field at all. Why not always use type, even in case the value is "online"?
According to the latest biblatex manual howpublished is supposed to be used for

howpublished field (literal)
A publication notice for unusual publications which do not fit into any of the common categories.

Altough I would not say that online publication are unsual publication, the type field on the other hand is supposed to be used for

type field (key)
The type of a manual, patent, report, or thesis. This field may also be useful for the custom types listed in § 2.1.3.

where under § 2.1.3 the custom types listed are artwork, audio, bibnote, commentary, image, jurisdiction, legislation, legal, letter, movie, movie, music, performance, review, standard, video, which is not really satisfying either.

🤔 What do you think?

[moewew]

See also #80 and #81.

I don't necessarily think howpublished is a particularly bad choice for the online designator and I would not immediately disregard it because of one word in the biblatex manual. After all the idea behind the howpublished field is clearly to give additional publication detail that did not fit elsewhere. The disadvantage is that the biblatex standard data model uses howpublished only for a few specific (indeed more unusual) entry types. I'm guessing that for the other types the existing fields + note/addendum would be enough to specify the relevant details.

type on the other hand is not the right field in my opinion. It clearly has a pre-defined role for @report and @thesis (@thesis+type = {phdthesis} is a PhD thesis, @thesis+type = {mathesis}, is a master's thesis etc.) that would clash with the online designator. (Plus, but this is more philosophical, 'type' feels like it should be about something inherent to the work and not something that is just incidental and related to the publication process.)

That all said, for the binary choice of whether or not you print the online designator it may very well make sense to define a per-entry boolean option. If you want a more general solution where users can give all sorts of medium designations for all entry types, you may want to define a completely new field mediumdesignator or some such.

[wg030]

Thank you very much for your detailed comment on that issue, @moewew

---

type on the other hand is not the right field in my opinion. It clearly has a pre-defined role for @report and @thesis (@thesis+type = {phdthesis} is a PhD thesis, @thesis+type = {mathesis}, is a master's thesis etc.) that would clash with the online designator. (Plus, but this is more philosophical, 'type' feels like it should be about something inherent to the work and not something that is just incidental and related to the publication process.)

That is pretty convincing to me and makes sense. Now that I think about it again I would esepcially agree to your remark in brackets that type should describe properties of the soruce rather than properties of the publication process. So I think we can indeed forget about type here.

---

The disadvantage is that the biblatex standard data model uses howpublished only for a few specific (indeed more unusual) entry types.

Well, this might be a reason to go for your suggestion to maybe add a new field mediumdesignator. But then again I would tend to make use of the standard data model as much as possible, but only if the intention of the standard model is not (completley) broken. So extending the use of howpublished in a quite complex style like the one we have would make sense from my point of view if the extension is not to far fetched from the standard model. But your

I don't necessarily think howpublished is a particularly bad choice for the online designator

suggests that this is the case here or did I misunderstand you?

---

That all said, for the binary choice of whether or not you print the online designator it may very well make sense to define a per-entry boolean option.

Yes, the more I think about that, I come to the same conclusion altough I must admit that the old version of the norm allows space for different interpretations.
Anyway, I am reading the new version of the norm from 2021 at the moment and I will try to focus on that subject, too. Since the new version is way more sophiscated I am confident that the possible interpretations will reduce and we can get a clearer picture of what would be best to do concerning the automatic adding of "online".

[moewew]

I guess the crux of the matter is simply that there is no field for a medium designator in the standard data model. Semantically, howpublished comes pretty close, but it is intended for a slightly different job (and it only applies to a certain subsets of entry types by default). Nevertheless, it is probably what I would use for an ad-hoc solution.

If you decide to use howpublished for the medium designator there is a chance of a "clash" (where a user needs both a medium designator and howpublished in its original capacity). Furthermore, there is the issue of portability of your .bib files. If you use howpublished for something that deviates from the standard data model, undesirable stuff may end up being printed with other styles.

A new field would be clear in its semantics, cannot clash with existing field uses and is unlikely to accidentally end up in the output with other styles. The usual disadvantage of new fields in contributed styles is that it can lead to an overall inconsistent data model if different styles use different names for essentially the same job. This is unlikely here, since medium designators are not that common. But you may want to liaise with the developer of biblatex-mla (https://github.com/jmclawson/biblatex-mla), which also has an option to output medium information.

[wg030]

I guess the crux of the matter is simply that there is no field for a medium designator in the standard data model.

🤔 Yes, it sound as if this is the exact reason that both ways have its advantages and disatvantages. So we have to decide which disatvanatges we can live with and which not.

---

If you decide to use howpublished for the medium designator there is a chance of a "clash" (where a user needs both a medium designator and howpublished in its original capacity). Furthermore, there is the issue of portability of your .bib files. If you use howpublished for something that deviates from the standard data model, undesirable stuff may end up being printed with other styles.

This is an extremely strong argument against howpublished and for the introduction of mediumdesignator, whereas

The usual disadvantage of new fields in contributed styles is that it can lead to an overall inconsistent data model if different styles use different names for essentially the same job.

is a disadvantage that I could live with. Hence I would even change my original tendency and vote for going with mediumdesignator now. But let us see what @michal-h21 and @DavidLuptak think about that.

---

But you may want to liaise with the developer of biblatex-mla (https://github.com/jmclawson/biblatex-mla), which also has an option to output medium information.

Thanks a lot for that hint. I haven't known that style yet and having a quick glance at it, it seems to be very sophsicated and we can learn a lot from his implementation not only concerning this question.
He seems to go with howpublished and a boolean package option for allowing the user to decide on their own if they want to display that information at all or not as far as this problem is concerned. Maybe a new package option might be a good choice for our package, too.