[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: IRC log - ODF Collab SC call - 2017-10-04
[16:34] Svante Schubert: Hi Patrick
[16:35] Svante Schubert: We have the ODF Plugfest next week and LibreOffice Conference
[16:36] Svante Schubert: I got two presentation next week for both sessions
[16:37] Svante Schubert: My pitch is taming ODF with better tooling on the ODF schema
[16:38] Svante Schubert: ODF is complex we have more than 600 elements and 1600 attributes, but even if we start with two elements alone, like text:h and text:p it is difficult without tooling
[16:38] Svante Schubert: If you and I would start over again writing an ODF applications, what would we do?
[16:39] Svante Schubert: How would we cope and shild ourselves from that complexity?
[16:39] Svante Schubert: Questions like what are the possible locations of these elements (text:h and text:p) and can they be nested?
[16:39] Svante Schubert: When do I have sufficient test documents to cover the common use cases?
[16:40] Svante Schubert: Or what are the difference between text:p and text:h :)
[16:40] Svante Schubert: So the idea is to add a better tooling on the ODF schema
[16:40] Svante Schubert: better automation, like mapping the schema into a graph database for analysis
[16:41] Svante Schubert: analysis or reverse engineering ;)
[16:41] Svante Schubert: Or annotation of logical blocks within the ODF schema
[16:42] Patrick: ODF as document interchange format versus ODF as document processing format.
[16:43] Svante Schubert: The pitch is not necessary targeting collaboration, but assistance for new ODF applications in general
[16:44] Svante Schubert: Simply put: We have a huge XML schema and it is hard to read, we might want to add better tooling to it. With additional future benefits, like automated ODF interoperability tests with test documents..
[16:47] Patrick: define regression?
[16:55] Svante Schubert: text:p might represent different features, like text:p within an image is a totally different feature that is not implemented by any ODF application
[17:00] Svante Schubert: we have to do the basic work to allow automation
[17:01] Patrick: For round-trip testing, use XQuery statement that tests for presence of element in the round-tripped document, boolean, T or F.
[17:01] Svante Schubert: like analysis of the ODF XML schema and annotiation of its logical blocks
[17:03] Svante Schubert: Jos did the XQuery approach on the ODF Server - https://gitlab.com/odfplugfest/odfserver
[17:03] Svante Schubert: I guess he is using XSLT as verification of the regression test (load/save roundtrip)
[17:04] Svante Schubert: The problem is that the XSL transformation have to be written manually
[17:05] Svante Schubert: I once wrote the XSL transformation from ODF to XHTML (the ODF specification is using)
[17:06] Svante Schubert: It took years in the end I have mixed three parts:
[17:06] Svante Schubert: 1) The business logic of ODF (what is a logical XML block and how it is represented
[17:06] Svante Schubert: 2) How the logical ODF block is being mapped to an HTML (mapping)
[17:06] Svante Schubert: 3) How the mapping had been implemented in XSLT
[17:07] Svante Schubert: ^^it was incredibly complex and should be separated, therefore I want to annotated 1) in the schema to base automation on it :)
[17:07] Patrick: Ok, what I meant was testing the structure of the resulting XML, which does not test the styles to be applied to it, doable with XQuery though.
[17:09] Svante Schubert: Jos tests everything style and structure with the ODF server
[17:09] Svante Schubert: The problem of the XQuery stuff is, that you can not reuse it for automated testing or change-tracking. Instead I would focus on the specification and implementation of changes (logical blocks of ODF XML and how they are being changed)
[17:12] Svante Schubert: In other words, I would not start with XQuery, but..
[17:13] Svante Schubert: I would try to generate XQuery from the annotation I made in the schema and other information we would need to add to the specification (e.g. what is the default attributes of a logical block, when you add a new one)
[17:16] Svante Schubert: Jos ODF server is only 'asking' if an ODF application is supporting a 'feature'
[17:16] Svante Schubert: by loading and saving the feature by the ODF application
[17:16] Svante Schubert: and checking if it is still checking afterwards
[17:16] Svante Schubert: via XSLT
[17:17] Svante Schubert: For instance, once a developer only checked that the formula attribute is still existing after the round-trip (load/save)
[17:17] Svante Schubert: not checking the content :/
[17:17] Patrick: by checking manually you mean it isn't generating every possible feature/context
[17:18] Svante Schubert: I mean they have written manually the XSLT to check for the certain features
[17:18] Svante Schubert: Once we had too many test documents not knowing what to use to test a feature, I fear soon we have too many tests in addition not knowing what we need to test a feature..
[17:19] Svante Schubert: they are not generating the XSLT from a feature description within the ODF specification :)
[17:20] Svante Schubert: its not easy, like the feature of page break on table (or paragraph) is quite often linked within XML files
[17:21] Svante Schubert: ^^I will take this as a home work :)
[17:22] Svante Schubert: how to specify the page break or pagination feature (switching from horizontal to vertical) for ODF
[17:23] Svante Schubert: With often linked, I mean: The paragraph is linked to the automatic style (in the content.xml) this one is linked to the styles.xml page style and the page style to another section where the values are written down
[17:23] Svante Schubert: Question: How to define this feature (easily readable by machines) to create an XSLT from it :)
[17:24] Svante Schubert: Thanks for joining! :)
[17:24] Svante Schubert: byebye
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]