Deduplication in Missa

[Michaelskapel]

In my effort to deduplicate the text items in Missa, I have compiled a redirection table that shows all possibilities to redirect a label in a certain code file to a label in a different code file. Almost all redirections (somewhere between 1500 and 2000!) have been implemented in the Nederlands version. Does it make sense to put the tabel (Redirection_Table.xlsx) and a description (Deduplication_in_Divinum_Officium_Missa.docx) on my Google drive and make it available to others?

[APMarcello3]

Please do!

[Michaelskapel]

Here are the links to Deduplication_in_Divinum_Officium_Missa.docx

https://docs.google.com/document/d/14OjoKWnH_Efi15_nDP7Qqg0PK27do6GU/

and to Redirection_Table.xlsx

https://docs.google.com/spreadsheets/d/14L5PNvKliQyTJmRPk4H7jAgm3rf20xk7/

The table, which is a work in progress, is without any warranty. I worked very carefully, but errors may still be present. Please double check before you implement a redirection: once the redirection is done, it is not easy to detect an error.

[mmolenda]

Hello, I'm not sure if it is directly related to this effort, but recently I spotted numerous refactoring and deduplication commits, mostly in scope of introduction of the Spanish language version by the author @d4h0y0sr. Some of the changes are not clear to me though. Could the author clarify several selected cases for me please?

• Why is Oratio section removed from web/www/missa/English/Sancti/01-01.txt while the parent proper is Sancti/12-25m3 which has a different Oratio? commit
• Why is Postcommunio removed from web/www/missa/English/Sancti/12-27.txt? commit
• Why is Postcommunio removed from web/www/missa/English/Commune/C2c.txt? commit

[Michaelskapel]

From the recent discussion, I have learned a lot about the Divinum Officium's label system. I want to check with you if my understanding of the system is correct.

1. There are four folders - Commune, Ordo, Sancti, and Tempora - which contain the code files.
2. The code files in the Latin version contain three types of labels, each with their own type of content: (i) text-labels with a (unique) text in Latin (ideally without duplications), (ii) reference-labels with a reference to another label (which may then contain text or another reference), and (iii) system-labels, such as [Rank] and [Rule], which can better be left unchanged in a vernacular version, except for obvious adaptations like the local names of Saints.
3. It is obvious that a vernacular version shall have the four folders and the location of the code files in sync with the Latin version.
4. The Latin version is leading, and any vernacular version inherits the label system from the Latin version.
5. A vernacular version has the (unique) text-labels in its own language; ideally, this is the only thing that needs to be provided to make a certain vernacular available in Divinum Officium.
6. Since a vernacular version inherits all the references from the leading version, there is - in principal - no need to keep reference-labels in the vernacular version; if the need of a reference-label still arises, such a label contains either (i) text or (ii) a reference to a label that contains this text.
7. Given the special status of the English version - which may take over the leading position from the Latin version - its (unique) text-labels shall be in complete sync with the ones in the Latin version.
8. The English version is discouraged to include labels that are not present in the Latin version; if such labels cannot be avoided, vernacular versions are forced to also keep such labels (with the appropriate content).

Comments are appreciated.

[d4h0y0sr]

1. Yes
2. Yes
3. Yes
4. Yes, in this case vernacular will inherit first from english then if empty from latin.
5. I think, i have never tried to add a new vernacular language from zero.
6. There should not be an scenario where a vernacular has reference but the latin does not has, that usually happens because english is obstructing the inheritance with unique text, the fix require to remove unique text from english, not to add reference label. Also notice we should avoid a reference label to another reference label, the reference label should be always to a text. Reference label that points to another reference label is known as "reference nesting" which is an issue and could generate problems.

7-8. I think so, i mean in terms of unique text (not references), every vernacular has to have the unique text, english does not matter in this scenario, as you have to write down the vernacular text either way. English just matters when you want to use references labels automatically from latin files without writting again these reference labels in vernacular files.

[Michaelskapel]

@d4h0y0sr : Thanks for you comments! A few remarks from my side:

4. I take it that by 'in this case' you actually mean 'in the present case'.

5. I did it for Nederlands! I took the Latin version as my template, but noticed that there were many text duplications; this was the reason for me to start constructing the redirection table (first to Commune and later to Tempora and Sancti). Having a completely deduplicated Latin version would make things much easier and less time-consuming (!) for new vernaculars.
   The Nederlands version is pretty unique at the moment (except for the combined labels like [Graduale], [Tractus], [GradualeP], etc.), but I had to keep reference-labels where the Latin version had a duplicated text-label. I do hope that the Latin version will be deduplicated soon, after which I can then remove all these no longer necessary reference-labels.

6. Note that the need for a reference-label still arises in the case of a reference to those Commune's Oratio, Secreta, and Postcommunio where 'N.' or 'N. et N.' has to be substituted by the proper vernacular name(s). It may be possible to solve this by defining the substitution in the [Rule] and adapt the program to take it from there, but I can live with the present situation.

7-8. You are right.
Since I did not feel authorized to touch the English version, I had to add the labels that were necessary to overcome the disturbed inheritance by the English version. But we really have to find a better solution: the program should be adapted such that the English version will never disturb the inheritance from the Latin version.

[d4h0y0sr]

4. In this case = any scenario where we expect vernacular to inheritance. Vernacular can not inheritance from latin if there is a section in english.
5. I suggest any translator to help in the Duplication efforts, there is no one in charge to do this task. If you have the missal, which is always the reference, you can be sure to fix the duplication taking into account to respect the accents and punctuation signs thats it, there are other developers that can review your changes in case you make mistakes. In that way, every "translator" would help not only to maintaining a language but also to keep latin in order.
6. You are totally right i forgot that scenario.
   7.-8. Given present software architecture of the website, i would say every translator is authorized to touch english just to remove inheritance obstruction, that benefits not only your vernacular language but all vernacular languages. Also its easier to catch mistakes in english as i would say is the most used language in the website.

[FAJ-Munich]

Maybe a few comments on the vernacular issue and your discussion which seems to be more pressing for the Missale than the Breviary at the moment:

1. Indeed English should not have any texts wherever a reference is used in Latin. However, pruning is harder done than said as – over time – diverging translations have cluttered in the database and the only authoritative source for decision making on the English, ultimately, is Fr. Marcello.
2. In general, the missa database is not as well kept than the horas unfortunately. Long-term, I hope the database can be merged such that every office/mass has its designated file and there are no divergences in [Rank] and Commemorations etc.
3. Within the [Rank] hash in the vernaculars, feel free to add your translations of the Feast and don't worry about anything past the first ;; group. It's ignored and taken from the Latin only.
4. Technically, there would be a hack to get a Language to be stacked on Latin (or any other vernacular) only: you would need to name it, e.g., Português/Latin-Portugues instead in the missa.dialog configuration. But be aware that, then, nothing is inherited from English anymore.
5. Whether a "check Latin for @-reference first" mechanism can easily be developed remains to be analysed.