CSV on the Web Working Group Teleconference

14 May 2014


See also: IRC log


Ivan Herman (Ivan), Phil Archer (phila), Jeni Tennison (JeniT), Jeremy Tandy (jtandy), Andy Seaborne (AndyS), David Ceolin (DavideCeolin), Jürgen Umbrich (jumbrich)
Dan, Eric, Yakov, Alfonso
Jeni Tennison
jtandy, JeniT


approve last weeks meeting record

<JeniT> RESOLUTION: accept previous meeting record at http://www.w3.org/2014/05/07-csvw-minutes.html

JeniT: aiming to publish another round of specs; please can the editors focus on what they need to do to get published

ucr document update

<jumbrich> ok, i use telephone on mac os , and every time the call gets cancelled after 15 minutes, anyway, i just redial in again

<JeniT> jtandy: we’ve added three new use cases

… one on RTL following discussion last week

jtandy: we need to arbitrarily say that the column with the identifier in could be anywhere

ivan: eh?

jtandy: normally, in CSV, the first column is the one that holds the identifier

… each row describes a thing

… the first column contains the identifier for the thing

… in RTL the identifier might be the other

<AndyS> and it may be composite ... several columns

… end, so identifiers might come from any column

… the other useful thing is that those with vertical writing still use columns in the same way

… the use case needs to be updated to say that

ivan: the complication is that there are languages where that isn’t a choice

jtandy: the important thing is that we have something in that recognises non-European ways of doing things

… we have the HL7 use case

… though it’s not a use case yet because it doesn’t declare any requirements

… it just says that the health community has a format

… it clearly provides examples of microsyntax for example

… but is that a use case

… there’s another format Yakov identified

… we could have CSV-type formats and match them against use cases

ivan: does the micro-syntax requirement come up in other places?

jtandy: yes

ivan: HL7 is obviously very important

… referring to it is good

JeniT: what about saying that a HL7 user wants to use generic CSV tools to validate it

jtandy: possibly, we need specific examples

ivan: we have someone on our team who is part of the Healthcare & Life Sciences interest group

… he’s heavily involved in HL7, so maybe we can ask him

… to provide a feasible story around it

<phila> HL7 Web site

jtandy: that would be great

ivan: I will contact him and introduce you

jtandy: use case #21 is another format-oriented use case

… it still needs to be user/action oriented

… on Darwin core, biodiversity group

… we’ve picked out the key aspects and we could create a story around it

… we’ll follow that up

… overall, there are some issues still to bottom out

<Zakim> phila, you wanted to talk about https://www.w3.org/2013/csvw/wiki/Use_Cases#Making_Sense_of_Other_People.27s_Data

phila: I’ve put a use case on the wiki at https://www.w3.org/2013/csvw/wiki/Use_Cases#Making_Sense_of_Other_People.27s_Data

… I hope that it can be included; it has real data

… and two new requirements

… one is third-party metadata (I describe someone else’s data)

… I’m happy to do more on it if that’s useful

… I’m also following up for verification

JeniT: I have a use case from ONS in which there are rows that would normally be columns

ivan: still no use case on XML conversion

phila: shall we ask Liam or Carine to come up with one?

ivan: I will do introductions

… in some way that’s more important, because if we’re driven by use cases then without one then we have no justification for an XML conversion

jtandy: I agree, we need such a use case in which it has to be converted to XML to enable their toolchain to operate on it

ivan: I think we already have made decisions on which requirements are accepted, deferred etc

… but this isn’t reflected in the document

DavideCeolin: I’ve started trying to reorganise them

… I’ve put the deferred requirements in a specific section

… most of the requirements were accepted aside from those that were incorporated into others, as I understand it

JeniT: I’d like to make a decision on the classification next week

DavideCeolin: I will send out an email to that effect

jtandy: the requirements are in some places skeletal, and we need to flesh them out

… I’d like to put discussion from the list into the document

<ivan> scribenick: jtandy

RDF generation

AndyS: talks about the text that Ivan has provided
... about the conversion process - doesn't fit that well into the metadata document
... would be better to fit into one document about conversions

ivan: best place would be the syntax doc

JeniT: or all three places :-)
... as long as it is somewhere this will be ok
... the location will "flush out" during review
... not clear if we need one over arching doc for all conversions

ivan: commented on the text copied in by AndyS ...
... in Ivan's original text he listed specific terms, but these are absent from the final text

<JeniT> http://w3c.github.io/csvw/metadata/#converting-tables

ivan: please can we raise this as a specific issue in the text

JeniT: "there might be some properties that need to be used in a consistent way to help the conversion"
... and which properties to use to annotate a CSV

<JeniT> http://w3c.github.io/csvw/metadata/#creating-annotated-tables

JeniT: thinking that section 2.1 of the metadata document is where I would talk about the association of metadata properties to rows/columns


JeniT: some stuff about creating the annotated table model, and some other stuff about the metadata for conversion

ivan: this is a secondary issue

JeniT: there's still lots of work to be done on the metadata docu
... asks AndyS if it would be useful to have a single templating mechanism for all conversions

AndyS: there would be different templates for each format
... the template provides a mechanism to map CSV to whatever format you need
... so each transformation (to a different format) would have a different template

JeniT: so if we had a requirement to convert to xml, would the template be XML or the "standard template format"

AndyS: probably ... we're trying to avoid needing to have a lot of tools to do the transformations
... there's no obligation to have complex tools
... the conversion can be done using lightweight text-based tools [I think that's what AndyS said]

<JeniT> http://golang.org/pkg/text/template/

JeniT: is there an existing text-template format (she Googles for knowlege!)
... can we propose a standard template form?

AndyS: suggesting that we start from URI Template

ivan: the "go-language" [?] you specify seems like to much
... it's got flow (if/then/else) in it
... let's keep it simple

AndyS: let's have the process "loop" driven by iterating through the rows ...

ivan: we should work out one conversion (e.g. RDF because we've already started)
... lets do JSON in parallel

AndyS: there might be slightly different rules (styles) of templating for converting to RDF or JSON

<JeniT> http://tools.ietf.org/html/rfc6570

phila: rfc talks about the different rules about what you can do in each part of the uri

AndyS: this is about the level of complexity as shell programming for substituting things based on the URI template

JeniT: please can we expose this on the email list?

AndyS: have done so already

JeniT: ah - but now I'm talking about "templating for all formats"

AndyS: ok .

ivan: ... only comment on csv2rdf doc
... restructure the RDF doc to be inline with the metadata doc - to help the general principles stand out

AndyS: what sort of restructuing are you thinking about

ivan: referring to the metadata entries specifically (as listed in the metadata doc) - to make the bridge between the two docs

AndyS: if we put the agenda out the day before the weekly call, this would help Gregg engage

JeniT: noted ... but the agenda is the same each week
... and takes an action to encourage people to propose things for the agenda

phila: agendas need to be out 24-hours before in order for formal resolutions to be passed

JeniT: ok - will work with danbri to publish agenda sooner & give people opportunities to add to the agenda

<phila> Eric S is on Gregg's time zone too - horrible for this slot :-(

JeniT: could find another time for the teleconf (to help US west-coast participation)

AndyS: there's always the tuesday slot - but we don't want duplication

model document

<JeniT> http://w3c.github.io/csvw/syntax/#bidirectionality-in-csv-files

JeniT: as agreed last week, the "embedded metadata" is now removed
... have been busy this morning

<AndyS> http://w3c.github.io/csvw/csv2rdf/#graph-template has a template example and refs RFC 6570

JeniT: section 4.4.2 gives the specific information about the model
... but there's an issue about whether we need a different media type for the different behaviours
... we need to resolve this issue

ivan: at the moment I disagree with specification in the model document; based around RTL behaviour

JeniT: see issue 13

ivan: ok - last comment not relevant ... will re-read

<JeniT> http://w3c.github.io/csvw/metadata/#bidirectional-tables

JeniT: will add some examples in here
... the RTL use case only has images

ivan: has the actual files - and will mail this around

JeniT: would be useful to have real examples

ivan: need to confirm with yakov about content of this file - because we don't know

JeniT: anything else to raise?
... AOB?


ivan: when do think (realisitically) we can publish again?

JeniT: 3-weeks

ivan: am in US in 2-weeks time, and w/c 2-June in Boston (should have time this week) other wise the 12th and 17th June could work

<JeniT> http://csvconf.com/

ivan: note the "information to editors" provided last time - if you do this it would help

JeniT: Notes "CSV Conf"

phila: should be able to go

JeniT: has submitted a paper

ivan: has everyone registered for TPAC

JeniT: (& phila) didnt know this was open yet
... CLOSE!

<ivan> trackbot, end telcon

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-05-14 13:36:24 $