- Tei-council - lists.tei-c.org

Datatype quandary
by Lou Burnard 26 Jun '15

26 Jun '15

We had agreed (fairly tentatively, but it is what Sebastian has since implemented) that in the new scheme of things, a purified attribute definition would simply embed a <dataRef> directly, rather than wrap it in some as-yet-unnamed alternative to the current <datatype> element. In testing this out against the Guidelines, I have now realised why this was not a good idea. Of the 466 instances of <datatype> in the current P5 source, a sizeable proportion (106) specify a @maxOccurs attribute to indicate that the attribute can (or cannot) take multiple instances of the datatype indicated. This can't be specified on <dataRef> itself at present, these attributes nopt being available. So we are left with the following choices: a) add @maxOccurs and @minOccurs to <dataRef> and cope with the ensuing chaos if the <dataSpec> being referenced permits multiple values b) define a whole raft of new <dataSpec>s and <dataRef>s, one for each combination of occurrence indicators and <dataref>, so that for example we will have teidata.pointer, teidata.pointers, teidata.atLeastTwoPointers etc etc. c) invent something which behaves just like <datatype> but has a different name and doesn't allow anything except <dataRef> as a child. d) permit <dataRef> as a child of <datatype> I bet smart chaps like you can tell which solution I am recommending... just for the record, the current model for <datatype> is an alternation of <macroRef> and "any-XML". In practice we use only the latter, as every <datatype> has a content resembling <rng:ref name="data.foo"/>. So proposal (d) above is arguably not really a change at all, since <dataRef> is just syntactic sugar for <macroRef> I'd like to move ahead on this this week, since it's holding up my testing of the purified version of the Guidelines. Can I ask for a quick response along the lines of EITHER "no problem" OR "hold on there I need to think about this" by Monday??? Or of course, "You idiot what about this completely different and much better solution"

2 4

Re: [tei-council] Datatype quandary
by Syd Bauman 26 Jun '15

26 Jun '15

I'm sure you've thought this through much more thoroughly than I, so I'm wondering if you can explicate this for me. Our end goal is to have a system in which the Guidelines, customization files, and non-TEI languages are expressed in PureODD, and all our processing chains work at least as well as they do now, right? But once we have the various processing working so that we can define one datatype using <dataSpec> and refer to it from a content model or attribute definition (or macro specification) with <dataRef>, why wouldn't we define them all that way? I realize that getting all our processing working is a tall order, as there is lots to do and some of it is very hard (or, at least, way beyond me). But if it works for one datatype, why not all? The answer might involve the fact that we're trying to build processing that can simultaneously handle *either* impure ODD or Pure ODD. While this seems nice to me, it doesn't strike me as a requirement. So perhaps I'm asking "why do we need co-existence?" If users are told "use all old or all new style ODD, submit old style to this process, new style to that process", I have no real problem with that. I guess I'm fuzzy on why co-existence is easier than big bang. If you could focus my lens on this, I'd appreciate it. > yes, if you want to go for co-existence you need all macroSpec and > new dataSpec coexisting in the same name pool. > > I dont think we are ready for a big bang conversion

1 0

Re: [tei-council] Touching Once (is so Hard to Keep)
by Syd Bauman 26 Jun '15

26 Jun '15

I think, or at least hope, this is the last of the "it didn't get delivered to list" re-posts. This one is from 2015-06-02 08:04, and was also sent to Sebastian, who answered me directly. Per his answer, reproduced below, I renamed the program 'Touch' to 'build-if-newer.sh' in r13246. (I post here in part because the commit message refers to this e-mail.) Since nothing has broken with the name-change (and some things work better), we could just delete the program entirely. > [Subject is the title of the last song on _Novella_, the 1977 > Renaissance album.] > > Can you tell me what the shell script "Touch" in directory P5/ is > for? It seems to do "If anything in ../P5/, ../Stylesheets/, or > ../I18N/ has been updated, then run a bunch of builds". But why, and > more importantly, why is it called "Touch"? Is it executed from some > other program(s) or only intended to be typed at the commandline? > > This turns out to be a real, if minor, problem, because while most > GNU/Linux systems default to case-sensitive file-systems, the Mac OS > X "Darwin" commandline defaults to a case-insensitive file-system. > Thus when running `make` on Mac OS X, this pgm gets executed when the > Makefile tries to execute a `touch`. > > I'd like to change the name of this cmd to > Touch.sh > build-if-updated.sh > make-if-updated.sh > or some such. > > P.S. I'm asking you, Sebastian, because no one but you has touched > this file (pardon the pun) in its entire Subversion hx on > current SF (i.e., since 2007-10 when you "put back executable > Touch"). SPQR> honestly, I have no memory of this script. i suggest you delete SPQR> it and see if anything breaks.

1 0

Re: [tei-council] bad date example
by Syd Bauman 26 Jun '15

26 Jun '15

This is from 2015-06-02 10:12, and is worth reading and reminding ourselves of the issue. > Indeed. Clearly there is controversy over this, although I thought it > would be pretty straigtforward. So that we don't forget to address > it, I've just created bug 762 and made it amber. > > BTW, yes, <precision> is, IMHO, the right answer. The problem there > is that @atLeast and @atMost are (appropriately) defined as > data.numeric, not a temporal datatype. > > Using LOC dates would also solve the problem. > > > Digging a little bit in the archives reveals that we had that sort > > of discussion before. Personally, I’m with Syd and would not mix > > notBefore/notAfter with from/to (nor with when) but this seems not > > to be agreed practice: > > * http://tei-l.970651.n3.nabble.com/Precision-of-time-information-for-state-t… > > * http://tei-l.970651.n3.nabble.com/datable-attributes-and-durations-of-uncer… > > BTW, for all sorts of fuzziness we have <precision>.

1 0

Re: [tei-council] New Jenkins build jobs for bleeding-edge and stable plugin
by Syd Bauman 26 Jun '15

26 Jun '15

This one from 2015-06-02 10:15. > > What do you see if you go to Help / Manage Add-ons? I see only one, > > the bleeding build. > > I see only one, but don't know which one it is. (Nor do I know how to > tell, unless I decipher the version #, which in this case is > "2.8.2".) > > > > I think the Preferences page just lists all updateSite.oxygen files > > that are known to Oxygen, not those which are active. > > > > Your post nicely copies my original post to the list, though, which > > I was supposed to re-send, so I'll consider it re-sent.

1 0

Re: [tei-council] repair 751, take 1
by Syd Bauman 26 Jun '15

26 Jun '15

I sent this 3 weeks ago (on 2015-06-05 23:32), but as far as I can tell, it never made it to the list itself. At least, it doesn't seem to be in the archives, and no one has responded. I realize some folks are off to my namesake city, but I'd like to get this off my plate. > Reminder to everyone to take a look at the solution in the branch at > https://svn.code.sf.net/p/tei/code/branches/P5/for-bug-751 Unless > there are objections, I'll plan to fold this in on or about Fri > 06-12. > > Problem: <event> is not allowed as a child of <org> > > Current ways <person> and <place> get child <event>: > > model.persEventLike => birth death event listEvent > model.placeEventLike => event > > > Proposed way <org>, <person>, and <place> get child <event> (in that > branch): > > new class, model.eventLike > > model.orgPart => model.eventLike > model.personPart => model.eventLike > place => model.eventLike > > removed model.placeEventLike and model.persEventLike (and made > <birth> and <death> directly members of model.personPart) > > > One might argue that we should keep model.persEventLike and > model.placeEventLike, and add model.orgEventLike. But since 2 of > those classes would have nothing but <listEvent> and <event> as > members, I took the slimmer route.

1 0

Fwd: Re: NVDL error from forums
by Martin Holmes 25 Jun '15

25 Jun '15

This is a patch from Radu. I've tested this and it works perfectly; our Spec and Guidelines files can now be validated. I'm waiting to hear if this is going to be part Oxygen 17.1. Either way, anyone editing specs or guidelines will probably want to install this. It seems to be the case that dataNode.xml, dataRef.xml, dataSpec.xml and interleave.xml are invalid, but I'm guessing it's because they reference each other, and can't validate against the last release of the schema. I'll try validating them against the latest Jinks NVDL. Cheers, Martin -------- Forwarded Message -------- Subject: Re: NVDL error from forums Date: Wed, 24 Jun 2015 09:19:02 +0300 From: Radu Coravu <radu_coravu(a)sync.ro> To: Martin Holmes <mholmes(a)uvic.ca> CC: Oxygen XML Editor Support <support(a)oxygenxml.com> Hi Martin, Applying the patch would be something like: 1) In the Oxygen installation folder there is a "lib" folder. In that folder you should create a new folder structure like: > endorsed/com/thaiopensource/validate/schematron/resources/ 2) In the "resources" folder copy the Schematron file I'm attaching. This patch should override the Schematron file which comes with the "jing.jar" library. Regards, Radu Radu Coravu <oXygen/> XML Editor, Schema Editor and XSLT Editor/Debugger http://www.oxygenxml.com On 6/23/2015 8:51 PM, Martin Holmes wrote: > Hi Radu, > > Many thanks for looking into the NVDL issue for us. On the board post: > > <http://www.oxygenxml.com/forum/topic11601.html> > > you mentioned a patch I could try out for you -- do you want to send it > over? > > The other solution (changing the prefix) is something we can fall back > to if we have to, but it will involve editing a lot of files (Schematron > is distributed all over our specification files), so I'd rather try > something else first. :-) > > Many thanks, > Martin >

1 1

interleave.xml?
by Martin Holmes 24 Jun '15

24 Jun '15

There's a file called interleave.xml in the Specs. Its log shows: r13221 | sbauman | 2015-05-27 18:03:01 -0700 (Wed, 27 May 2015) | 1 line Add the 'Id' Subversion keyword to source files that don't have it. ------------------------------------------------------------------------ r13076 | louburnard | 2014-11-21 04:18:32 -0800 (Fri, 21 Nov 2014) | 2 lines add interleave; fix bug #700 in tite Is it still an active part of the Pure ODD plan? Cheers, Martin

2 1

odd gotcha (or bug?)
by Lou Burnard 24 Jun '15

24 Jun '15

ODD processing always has a surprise up its virtual sleeve. Let's suppose your ODD wants to add <persName> and <orgName> and <placeName> from the namesdates module. Let's further suppose you want to see only W3C date values in your dating attributes. So your ODD has something like this <moduleRef key="namesdates" include="orgName persName placeName"/> <classSpec type="atts" ident="att.datable-iso" mode="delete" module="namesdates"/> <classSpec type="atts" ident="att.datable-custom" mode="delete" module="namesdates"/> Sound plausible? But it doesn't do what you expect. In the first place, the presence of the module attribute on <classSpec> generates the following entirely mysterious schematron error : E [ISO Schematron] Specification att.datable-custom: the value of the module attribute ("namesdates") should correspond to an existing module, via a moduleSpec or moduleRef In the second place, the moduleRef to namesdates will naturally include those classes along with the elements you've requested. In the third place, if you realise this is happening (and you are a smart ODD coookie if you do), you can replace the moduleref with elementRefs for the elements you want <elementRef key=orgName/> <elementRef key="persName"/> <elementRef key="placeName"/> Whereupon, huzza, those pesky classes disappear. Not because of your <classSpec>s of course, but because you've removed the moduleRef` In fact you can remove the classSpecs completely: if there is any way to delete a class which is specific to a module (i.e. not defined in the tei module), this isn't it, though it works fine for classes which *are* defined in that module.

4 16

pics from Ann Arbor
by Syd Bauman 23 Jun '15

23 Jun '15

No time to do anything nice with these, so this is just a pile (of reduced-size images) without any digital editing or album software. These two links are to the same set of 52 images: http://bauman.zapto.org/~syd/temp/4TEICouncil/Ann_Arbor_2015-05/ http://bauman.zapto.org/~syd/temp/4TEICouncil/Ann_Arbor_2015-05.tgz The all-at-once tarball is ~11 MiB.

1 0

NVDL validation of spec files
by Martin Holmes 23 Jun '15

23 Jun '15

As you may have noticed, our NVDL validation process for files in the P5 Source has been failing recently, with this error: "Attribute @xsl:dummy-for-xmlns is not allowed on element <xsl:stylesheet>" I raised this on the Oxygen forum, and we have two possible solutions suggested by Radu: <http://www.oxygenxml.com/forum/post34459.html> One involves a patch to Oxygen, which I'm going to try out first. If that works, I think we should just wait for it to come out in 17.1. The other involves changing the namespace prefix in our isosch file, which looks trivial but will need some investigation to make sure it will work just in that individual file, or whether there are other uses of that prefix that will need to be brought in line elsewhere in the specs. I don't think so (except for one in constraintSpec.xml), but some caution might be in order; there's a lot of stuff in Utilities that looks like it might be affected. Cheers, Martin

1 0

Re: [tei-council] Three kinds of data ref
by Lou Burnard 19 Jun '15

19 Jun '15

But why would you bother to do that when you can put a valList in dreckly? (could use xinclude if v lazy) On 19/06/15 18:56, Sebastian Rahtz wrote: > @ref could point to a remote TEI doc with a dataSpec which is just a valList > > Sebastian

1 0

dataSpec content
by Lou Burnard 12 Jun '15

12 Jun '15

I had proposed defining dataSpec without any wrapper element for the parts of it which actually define the datatype itself i.e. with an (alternate|dataRef|valList) in its content model. Sebastian points out, reasonably enough, that this makes life difficult for someone wishing to modify or change the "content" of an existing dataSpec, and is not consistent with our previous practice. I take the point, but find it hard to imagine anyone wanting to write an ODD which would modify one of our existing datatypes, when defining a new one with a different name would be much simpler and more transparent. Indeed, maybe you shouldnt be allowed to modify existing datatypes. Thanks to Sebastian's much-appreciated late night activity, the Pure ODD branch on SF (P5-Pure) now has a test file and we have a test implementation to test.... yours in purity L

3 2

p5odds-examples-odd
by Lou Burnard 10 Jun '15

10 Jun '15

I just noticed that recent changes have made p5odds-examples.odd invalid. This is the ODD file that defines the schema which is used by nvdl to validate examples in the body of the Guidelines, so it's fairly crucial that it continue to function. The error is only detected by schematron, and relates to defaultVal fiddling: I have corrected the ODD and checked it in, but I am now wondering at what stage exactly in the build process does the corresponding RNG file get generated. Simply running Make doesn't hack it, so it must be amongst the arcana involved in doing a new distribution. anyone else noticed this?

2 3

F2F Minutes Frozen
by Hugh Cayless 10 Jun '15

10 Jun '15

I’m getting ready to publish the minutes from Ann Arbor, so I have frozen the Google Doc. If you have any remaining changes you want to put in, you should let me know immediately. Thanks, Hugh /** * Hugh A. Cayless, Ph.D * Chair, TEI Technical Council * Duke Collaboratory for Classics Computing (DC3) * hugh.cayless(a)duke.edu * http://blogs.library.duke.edu/dcthree/ **/

1 0

Re: [tei-council] dataypes: the never ending story
by Syd Bauman 09 Jun '15

09 Jun '15

I understand completely that conversion to Pure ODD is a big deal, but is converting <datatype> to <dataSpec> going to have a much larger impact than using <elementRef> instead of <rng:ref>, etc.? My point is that conversion to Pure ODD is a big deal either way, no? I think you're right, project management, and maybe even some money, is in our future. (And I can already hear Hugh muttering "git-fork" under his breath. :-) > my heart slightly fails me when I start thinking about the side > effects of dropping <datatype> and converting all the Guidelines to > <dataSpec>. The existing .odd files out there stop working > (potentially), Roma stops working, as does its little cousin > Byzantium, the documentation of the Guidelines probably gets hosed, > a small myriad of utilities fall over. > > I hate to say it, but changes of this scale need a bit of project > management. > > else you need a team of dedicated people fixing things on the fly.

2 1

dataypes: the never ending story
by Lou Burnard 09 Jun '15

09 Jun '15

If we are to abide by Martin's eminently suggestion of permitting the new pure odd dataRef to coexist with the current <dataype> mechanism, I can't see any way out of doing the following a) make a set of new <dataSpec> elements corresponding with the existing <macroSpec type="dt">s b) since @ident values are supposed to be unique, these new things will all have to have different names (I suggest teidata.foo rather than data.foo) c) the current section 1.4.2 on datatype macros (#DTYPES, in #ST) will also need to be cloned, or substantially revised, since this is where the pesky things are actually defined Anyone got any better ideas?

3 11

unencoded words that maybe should be
by Syd Bauman 08 Jun '15

08 Jun '15

I'm going over the minutes, and strolled across bug 740. It lists an instance where a few words that should have been encoded as <gi> were not. I wondered about others. So I created http://bauman.zapto.org/~syd/temp/4TEICouncil/list-of-PCDATA-idents.html In principle, it is probably a useful task for someone to go through this set and find the few that really should be encoded as <gi> or <ident> and fix them. In fact, it's obviously a huge task that would take multi-person coordination and lots of time for little benefit.

1 0

iso schemaron syntax
by Lou Burnard 08 Jun '15

08 Jun '15

Does this look vaguely right <sch:assert test="name(parent(.))='dataSpec' and not(contains(dataRef))">An alternate element inside a dataSpec may contain only dataRef children </sch:assert> ? #tooLazyToLookItUp

2 2

oxygen-tei plugin
by Martin Holmes 08 Jun '15

08 Jun '15

Hi all, Sebastian released a new version of the Stylesheets yesterday, prompting a new version of the oxygen-tei plugin, so I've taken the opportunity of this release to enact some changes we agreed on a while ago: The new version number of the plugin is 3.0.0; this disentangles its version numbering from P5 releases. The new location and file name for the plugin is: http://sourceforge.net/projects/tei/files/oxygen-tei/oxygen-tei-3.0.0.zip The file is now named oxygen-tei-x.x.x.zip to match the actual project name and to distinguish it from the debian package, and it's in a new folder called "oxygen-tei". I've left the old "tei-oxygen" folder on SF in place for the moment, just in case, so right now there's additional potential confusion, but if all is well after a couple of days I'll do two things: 1. Move the last couple of releases with the old filenames into the new folder and give them new filenames, adjusting the updateSite.oxygen file accordingly. 2. Remove the old folder and the rest of the old versions. Please yell if you don't think I should do either of those things. Cheers, Martin

2 2

Implementation of @selector on <rendition>
by Martin Holmes 08 Jun '15

08 Jun '15

At the FtF I got the go-ahead to implement the feature request described in detail here: <http://wiki.tei-c.org/index.php/Proposal_for_@selector_attribute_on_renditi…> with the proviso that I should once more call it to everyone's attention for any final objections before I go ahead and do it. If I don't hear objections in the next few days I'll make this change. Cheers, Martin

1 0

nvdl validation of Spec files in Oxygen not working
by Martin Holmes 08 Jun '15

08 Jun '15

Lou and I discovered the other day that the current setup by which Oxygen validates Spec files when you're editing them does not work for some reason. That means that when you're editing a Spec file, you have to be extra careful not to screw anything up. We don't know why it's not working; it looks like something changed around the time of the release of Oxygen 16.1. Cheers, Martin

1 0

defaultVal deprecation
by Martin Holmes 08 Jun '15

08 Jun '15

Hi all, I think a couple of messages got lost when the lists went weird last week, so I thought I should summarize something I need feedback on. I was under the impression that I had the go-ahead to remove defaultVals when the Council was convinced it was a good idea, but Syd reminded me that we should actually deprecate things for a couple of years before removing them. However, there is no mechanism for deprecating <defaultVal>; it's not a member of att.deprecated so it can't have @validUntil added to it. So I hereby recommend that <defaultVal> be added to att.deprecated. If I don't hear any shouts of objection within the next couple of days I'll do that. That of course will require that the Guidelines build then do something with @validUntil on <defaultVal>. That will be my next order of business on this ticket. Meanwhile, I've created this wiki page with all the current defaultVals: <http://wiki.tei-c.org/index.php?title=DefaultVal_candidates_for_removal> It turns out that there are no required attributes with defaultVal, so that low-hanging fruit doesn't exist. Cheers, Martin

1 0

New Jenkins build jobs for bleeding-edge and stable plugin
by Martin Holmes 04 Jun '15

04 Jun '15

Hi all, I have what looks like a working setup on my Jenkins box for building two versions of the Oxygen TEI plugin. The original build job, "oxygen-tei", to which some of you may have subscribed, is now replaced by "oxygen-tei-bleeding", to which you can subscribe here: <http://teijenkins.hcmc.uvic.ca/job/oxygen-tei-bleeding/lastSuccessfulBuild/…> while there is also now "oxygen-tei-stable", to which if you want you can subscribe here: <http://teijenkins.hcmc.uvic.ca/job/oxygen-tei-stable/lastSuccessfulBuild/ar…> The idea is that oxygen-tei-bleeding builds with the latest plugin code against the latest trunk builds of P5 and the Stylesheets, enabling us to see if any change to any of these has broken the plugin; oxygen-tei-stable builds with the latest plugin code, but against the current release versions of P5 and the Stylesheets, enabling us to see what a release version of the plugin would look like if either we or Syncro built one right now. The build is now being done with ant, and in addition to these two targets, there's also: ant syncro which builds a version for the Syncro folks using the latest stable P5 and Stylesheets, as they would do when creating a build for a new Oxygen release. Up to now they've been using our original update-and-upload.sh script, and then preventing it from uploading to SourceForge as it tries to do. That script also requires that you _know_ the correct versions of P5 and the Stylesheets, and supply them correctly on the command line; the new ant build script works that information out for itself. Finally, there is: ant stable which is what our release technician would use during a P5 release process, or what we would use to release an official build of our version of the plugin. This builds in the same way as the version above, but also offers to upload the result to SourceForge, and creates a new version of the updateSite.oxygen file which it offers to upload to tei-c.org. This should make releasing a new plugin version much easier, I hope (although it can only truly be tested for real when we come to release a version, so I'd like to be the one who runs it for real the first time in case something goes wrong). I've added the new job configs for Jenkins to SVN, where James can find them for the Oxford Jenkins, when he has the time. This is all brand-new and only gently tested, so let me know if you try anything, and please report any problems with as much detail as you can. Cheers, Martin

3 4

Minutes Doc and scheduling
by Hugh Cayless 04 Jun '15

04 Jun '15

Hope that by now everyone’s made it home safely. Can we all look over the minutes doc while the meeting is still fresh in our memories and correct/add anything that’s missing? I will plan to freeze the document on Friday June 5th and convert the minutes over for the website. Re scheduling the next telecon: Do we want to try to stick with Fridays at 09:00 or 10:00 EDT? That’s 06:00/07:00 PDT, 14:00/15:00 BST, 15:00/16:00 CEST. I’m flexible as to day of the week during the summer—when I’m not actually away, of course. Fridays in June/July look particularly difficult. I’ll either be on vacation or away on business for several of them. Doodle poll for June telecon: http://doodle.com/q5ixp76erdceeydk <http://doodle.com/q5ixp76erdceeydk> Doodle poll for July telecon: http://doodle.com/xy2hqrpk9swnzhzd <http://doodle.com/xy2hqrpk9swnzhzd> Doodle poll for August telecon: http://doodle.com/txi5p2cn3mhmfizr <http://doodle.com/txi5p2cn3mhmfizr> Thanks to all of you for your hard work last week. It felt to me like we covered a lot of ground. All the best, Hugh

8 10