1 | <!-- General principles |
---|
2 | |
---|
3 | * Where the Archimedes DTD is good, let's stick with it. |
---|
4 | * Conform to TEI when we don't have a reason for not conforming. |
---|
5 | ** So much of TEI Lite will be implemented |
---|
6 | * Use W3C standards (xml:id, xml:lang, xlink) where available and |
---|
7 | applicable -- rather than corresponding things in TEI/Archimedes |
---|
8 | * Allow for import of other useful vocabularies (MathML, etc.) |
---|
9 | * If we want want other features for e.g. special manuscripts or genres, |
---|
10 | let's use separate namespaces. |
---|
11 | * XHTML may provide good list and table models -- they are powerful |
---|
12 | and well-known |
---|
13 | * Base metadata on DC |
---|
14 | * Grammar will be maintained in Relax NG -- which can be converted to |
---|
15 | other formats as need with trang |
---|
16 | ** allows modularity |
---|
17 | ** allows datatyping |
---|
18 | ** lighter weight than XSD |
---|
19 | * Perhaps use Steve Tinney's literate Relax NG tools, which allow for |
---|
20 | automatic generation of ((X)HTML) documentation, etc. |
---|
21 | --> |
---|
22 | |
---|
23 | <echo xml:lang="la"> |
---|
24 | <info xmlns:dc="http://purl.org/dc/elements/1.1/" |
---|
25 | xmlns:dcterms="http://purl.org/dc/terms/"><!-- rename metadata? --> |
---|
26 | <!-- use DC here, where appropriate --> |
---|
27 | <!-- all the TEI header stuff seems needlessly complex --> |
---|
28 | <dc:title></dc:title> |
---|
29 | <dc:creator></dc:creator> <!-- clarify principles --> |
---|
30 | <dc:subject></dc:subject><!-- probably not so useful for us --> |
---|
31 | <dc:description></dc:description> |
---|
32 | <dc:publisher></dc:publisher> |
---|
33 | <dc:contributor></dc:contributor> |
---|
34 | <dc:date></dc:date><!-- this should be in a standard format --> |
---|
35 | <dc:type></dc:type> |
---|
36 | <dc:format></dc:format> |
---|
37 | <dc:identifier></dc:identifier> <!-- this is the locator! --> |
---|
38 | <dc:source></dc:source> |
---|
39 | <dc:language s:type="t:ISO639-2">akk</dc:language> <!-- ISO 639 --> |
---|
40 | <dc:relation></dc:relation> <!-- ??? --> |
---|
41 | <dc:coverage></dc:coverage> <!-- ??? --> |
---|
42 | <dc:rights></dc:rights> <!-- this is important! --> |
---|
43 | |
---|
44 | <!-- useful qualified DCMES --> |
---|
45 | <dcterms:created></dcterms:created> <!-- date --> |
---|
46 | <dcterms:modified></dcterms:modified> <!-- date --> |
---|
47 | <dcterms:copyrighted></dcterms:coprighted> <!-- date --> |
---|
48 | |
---|
49 | <!-- add reference to external metadata here --> |
---|
50 | </info> |
---|
51 | |
---|
52 | <text><!-- as in TEI --> |
---|
53 | |
---|
54 | <front><!-- as in TEI; optional --> |
---|
55 | </front> |
---|
56 | |
---|
57 | <body><!-- as in TEI --> |
---|
58 | |
---|
59 | <!-- recommendation: Archimedes chap was unfortunate, since it led to |
---|
60 | artificial subchap1, subchap2, etc.; at the same time, I propose |
---|
61 | that we eschew TEI <div0>-<div7> and instead use a single HTML-like |
---|
62 | <div> with an attribute, e.g.: |
---|
63 | |
---|
64 | <div level="1" type="chapter" n="1" xml:id="N123456"> |
---|
65 | |
---|
66 | <div level="1" type="canto" n="5" xml:id="N234567"> |
---|
67 | |
---|
68 | Let's keep n automatically numbered, as in CDLI. But let's add |
---|
69 | an attribute ref, on which we can place *conventional* reference |
---|
70 | designations, if they exist (e.g., Stephanus numbers for Plato, |
---|
71 | Bekker numbers for Aristotle, etc.). Let's allow ref at any level |
---|
72 | where it might reasonably be useful, and make it optional. |
---|
73 | |
---|
74 | Let's allow head freely, as in TEI, for short headings |
---|
75 | (non-sentences), which are containers, as e.g. "Theorem 1": |
---|
76 | |
---|
77 | 2007-04-03 Needs to be clarified: is there one head that |
---|
78 | specifies the name of div? |
---|
79 | |
---|
80 | <head>Theorem 1</head> |
---|
81 | |
---|
82 | Might as well allow also TEI trailer, no? |
---|
83 | |
---|
84 | To be deferred for the moment: conventions for verse & dramatic |
---|
85 | genres (lines, stanzas, speaker, etc.). This should be modularized |
---|
86 | and put into a separate namespace, anyway. |
---|
87 | |
---|
88 | pb and lb do not need fundamental reconsideration. |
---|
89 | |
---|
90 | 2007-04-03: (rc) images associated by link with pb |
---|
91 | 2007-04-03: (dw) NB: Brian's viewer requires at least one pb in a doc |
---|
92 | |
---|
93 | We keep emph as in Archimedes (no TEI hi), but make it a real element; |
---|
94 | not a milestone. (This was a mistake in Archimedes.) In addition, |
---|
95 | emph is also an attribute available at higher levels. |
---|
96 | |
---|
97 | Keep foreign, as in TEI & Archimedes, but with standard |
---|
98 | xml:lang. |
---|
99 | |
---|
100 | 2007-04-03: how to indicate non-standard encodings (e.g. Beta |
---|
101 | Code) |
---|
102 | |
---|
103 | We don't use mentioned, term, or title, as we normally don't |
---|
104 | mark these, and want to allow for doing so out-of-band. They can |
---|
105 | be included in a module, if someone wants them. |
---|
106 | |
---|
107 | What about q? 2007-04-03: consensus is incorporate; open & close |
---|
108 | attributes; allow rendering to default based on language |
---|
109 | |
---|
110 | <q type="inline" open="xxx;" close="xxx;"/></q> |
---|
111 | |
---|
112 | 2007-04-03 (rc): what to do about asymmetry? |
---|
113 | |
---|
114 | TEI note is a good idea, rather than <p type="margin">, etc.; let |
---|
115 | its attributes remain a matter for further study. |
---|
116 | |
---|
117 | TEI cross refs & links will not be used; instead we use xlinks. |
---|
118 | Everywhere. |
---|
119 | |
---|
120 | 2007-04-03 (dw): we may need to add our own attributes |
---|
121 | |
---|
122 | Editorial interventions: let's use reg ("regularized") for both |
---|
123 | abbreviation expansion and error correction: |
---|
124 | |
---|
125 | <reg orig="Bd." type="abbrev">Band</reg> |
---|
126 | <reg orig="misrpinted" type="error" resp="paul">misprinted</reg> |
---|
127 | |
---|
128 | TEI orig we don't need; I've chosen to merge corr with reg and ignore |
---|
129 | sic. |
---|
130 | |
---|
131 | TEI add and del are a good idea for MSS. |
---|
132 | |
---|
133 | gap we already have. |
---|
134 | |
---|
135 | Names, dates, and other entities we mark out of band. |
---|
136 | |
---|
137 | We use TEI num to mark numbers as non-linguistic elements of the |
---|
138 | text and also indicate their Arabic normalization (useful so they |
---|
139 | don't get submitted for morphological analysis): |
---|
140 | |
---|
141 | <num value="33">xxxiii</num> |
---|
142 | |
---|
143 | I also recommend var: |
---|
144 | |
---|
145 | triangulum <var>ABC</var> |
---|
146 | |
---|
147 | abbr to be merged with reg |
---|
148 | |
---|
149 | I suggest importing lists and tables from XHTML. |
---|
150 | |
---|
151 | figure is more or less OK. |
---|
152 | |
---|
153 | We keep s as "basic semantic unit"; not "orthographic sentence", as in |
---|
154 | TEI. |
---|
155 | |
---|
156 | something like TEI seg, HTML span (all-purpose)? |
---|
157 | |
---|
158 | 2007-04-03: consensus is against the preceding |
---|
159 | 2007-04-03 (dw): how to handle subscript/superscript |
---|
160 | |
---|
161 | formulas to be encoded with MathML, etc. |
---|
162 | |
---|
163 | container element <alien-data></alien-data> |
---|
164 | |
---|
165 | What to do about entities? maybe define <entity> tag |
---|
166 | --> |
---|
167 | |
---|
168 | <!-- keep p, as in TEI & archimedes --> |
---|
169 | |
---|
170 | </body> |
---|
171 | |
---|
172 | <back><!-- as in TEI; optional --> |
---|
173 | </back> |
---|
174 | |
---|
175 | </text><!-- as in TEI --> |
---|
176 | </echo> |
---|
177 | |
---|
178 | <!-- open issues: |
---|
179 | |
---|
180 | allow composite texts; something like group? |
---|
181 | I don't think we need anything like TEI corpus |
---|
182 | |
---|
183 | --> |
---|