| 1 | <!-- |
|---|
| 2 | //----------------------------------------------------------- |
|---|
| 3 | // Copyright Christian Arnault LAL-Orsay CNRS |
|---|
| 4 | // arnault@lal.in2p3.fr |
|---|
| 5 | // See the complete license in cmt_license.txt "http://www.cecill.info". |
|---|
| 6 | //----------------------------------------------------------- |
|---|
| 7 | --> |
|---|
| 8 | |
|---|
| 9 | <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN"> |
|---|
| 10 | <html> |
|---|
| 11 | <head> |
|---|
| 12 | <title>FAQ on CMT</title> |
|---|
| 13 | </head> |
|---|
| 14 | |
|---|
| 15 | <body> |
|---|
| 16 | <center><h1>FAQ on CMT</h1></center> |
|---|
| 17 | <h1>Index</h1> |
|---|
| 18 | |
|---|
| 19 | <ol> |
|---|
| 20 | <li><a href="#Providing sourceless documents">Providing sourceless documents</a> |
|---|
| 21 | <li><a href="#Accessing packages in a sub-directory">Accessing packages in a sub-directory</a> |
|---|
| 22 | </ol> |
|---|
| 23 | |
|---|
| 24 | <hr> |
|---|
| 25 | <h1><a name="Providing sourceless documents"/>Providing sourceless documents</h1> |
|---|
| 26 | |
|---|
| 27 | <blockquote> |
|---|
| 28 | <i> |
|---|
| 29 | I'm always trying to use CMT together with java and I now have |
|---|
| 30 | problems with the documentation. |
|---|
| 31 | |
|---|
| 32 | <p> What I'd like to do is just generating automatically my |
|---|
| 33 | documentation using javadoc. Since it is not done in cmt, I tried to |
|---|
| 34 | write a new fragment to define a new document type : javadoc. |
|---|
| 35 | |
|---|
| 36 | <p>The problem is that the process of generating the doc is not a |
|---|
| 37 | file to file process : it does not take a .java file to write a |
|---|
| 38 | .html. It takes a java package name (which has no associated file) |
|---|
| 39 | and returns many files. This is not possible with the current |
|---|
| 40 | implementation of document. It seems that adding this feature would |
|---|
| 41 | lead to write some code into cmt_generator. |
|---|
| 42 | |
|---|
| 43 | <p>Does anyone see another method to generate my documentation ? |
|---|
| 44 | |
|---|
| 45 | <p>Sebastien |
|---|
| 46 | </i> |
|---|
| 47 | </blockquote> |
|---|
| 48 | |
|---|
| 49 | <p>In fact what you ask for is already available in CMT, it is |
|---|
| 50 | documented (although likely crypticly ;-) ) as follows : |
|---|
| 51 | |
|---|
| 52 | <p>Since your "document generator" does not make use of any source |
|---|
| 53 | file per se, it's likely that you will provide as the fragments the |
|---|
| 54 | following files (I use as an example the name "javadoc" for your |
|---|
| 55 | document type): |
|---|
| 56 | |
|---|
| 57 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 58 | <b> |
|---|
| 59 | <pre> |
|---|
| 60 | ... |
|---|
| 61 | make_fragment javadoc_header |
|---|
| 62 | make_fragment javadoc -header=javadoc_header |
|---|
| 63 | ... |
|---|
| 64 | </pre> |
|---|
| 65 | </b> |
|---|
| 66 | </font> |
|---|
| 67 | |
|---|
| 68 | <p> where javadoc is simply ... empty. |
|---|
| 69 | |
|---|
| 70 | <p>I remind you that here : |
|---|
| 71 | |
|---|
| 72 | <ul> |
|---|
| 73 | <li> javadoc_header will be instanciated once per document |
|---|
| 74 | <li> javadoc will be instanciated once per source file (thus never if |
|---|
| 75 | no source file is specified) |
|---|
| 76 | </ul> |
|---|
| 77 | |
|---|
| 78 | <p>Then a document associated with the javadoc type could be specified |
|---|
| 79 | as follows, using the <i>unusual</i> feature I mentioned above |
|---|
| 80 | |
|---|
| 81 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 82 | <b> |
|---|
| 83 | <pre> |
|---|
| 84 | document javadoc MyDoc JAVAPACKAGE=MyPackage |
|---|
| 85 | </pre> |
|---|
| 86 | </b> |
|---|
| 87 | </font> |
|---|
| 88 | |
|---|
| 89 | <p>Here the syntax <b>JAVAPACKAGE=MyPackage</b> corresponds to |
|---|
| 90 | providing user-defined templates in the document's fragments |
|---|
| 91 | |
|---|
| 92 | (see the general <a href="CMTDoc.html#kw-constituent-option">syntax</a>) |
|---|
| 93 | |
|---|
| 94 | every variable name should then correspond within the fragment to |
|---|
| 95 | constructs like (appearing anywhere in the fragment) |
|---|
| 96 | |
|---|
| 97 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 98 | <b> |
|---|
| 99 | <pre> |
|---|
| 100 | ... |
|---|
| 101 | ... ${JAVAPACKAGE} ... |
|---|
| 102 | ... |
|---|
| 103 | </pre> |
|---|
| 104 | </b> |
|---|
| 105 | </font> |
|---|
| 106 | |
|---|
| 107 | <blockquote> |
|---|
| 108 | <i> |
|---|
| 109 | (please note the <b>${..}</b> syntax with required braces around the |
|---|
| 110 | variable name) |
|---|
| 111 | </i> |
|---|
| 112 | </blockquote> |
|---|
| 113 | |
|---|
| 114 | <p>You have probably already understood that CMT provides internal |
|---|
| 115 | predefined similar <a href="CMTDoc.html#Standard templates for makefile fragments">variable names</a> |
|---|
| 116 | |
|---|
| 117 | <p>the typical example being <b>CONSTITUENT</b> which is replaced by the |
|---|
| 118 | constituent name when the fragment is instanciated. |
|---|
| 119 | |
|---|
| 120 | <p>What I'm talking about here consists in adding any number of |
|---|
| 121 | user-defined variables which value will be specified on the <b>document</b> |
|---|
| 122 | statement line using any number of <b><variable-name>=<value></b> constructs. |
|---|
| 123 | |
|---|
| 124 | <p>You will notice that the all-uppercase convention has been selected |
|---|
| 125 | for CMT-internal variables (eg. <b>CONSTITUENT</b>). This is why I have used |
|---|
| 126 | this convention for this example. This is only a convention, any form |
|---|
| 127 | is possible here. The only constraint being that it should not overlap |
|---|
| 128 | with a macro name or an env. var name. |
|---|
| 129 | |
|---|
| 130 | <hr> |
|---|
| 131 | <h1><a name="Accessing packages in a sub-directory"/>Accessing packages in a sub-directory</h1> |
|---|
| 132 | |
|---|
| 133 | <blockquote><i> |
|---|
| 134 | <p>I tried to use CMT with sub-packages. How does this work? If I |
|---|
| 135 | have on cvs a structure like |
|---|
| 136 | |
|---|
| 137 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 138 | <b> |
|---|
| 139 | <pre> |
|---|
| 140 | Velo/VeloEvent/.. |
|---|
| 141 | /VSicbCnv/.. |
|---|
| 142 | </pre> |
|---|
| 143 | </b> |
|---|
| 144 | </font> |
|---|
| 145 | |
|---|
| 146 | <p>and on disk: |
|---|
| 147 | |
|---|
| 148 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 149 | <b> |
|---|
| 150 | <pre> |
|---|
| 151 | Velo/VeloEvent/v3/.. |
|---|
| 152 | /VSicbCnv/v3/.. |
|---|
| 153 | </pre> |
|---|
| 154 | </b> |
|---|
| 155 | </font> |
|---|
| 156 | |
|---|
| 157 | <p> it recognizes the stuff in the use statement, but when I say |
|---|
| 158 | <b>cmt show packages</b> they are not there... Hence, the |
|---|
| 159 | add-in for devstudio fails. |
|---|
| 160 | </i></blockquote> |
|---|
| 161 | |
|---|
| 162 | <p>Well this is just a subtle effect : |
|---|
| 163 | |
|---|
| 164 | <ul> |
|---|
| 165 | <li> when you are in the context of say <b>VeloEvent/v3</b>, |
|---|
| 166 | then the path of the current package is implicitly added to the |
|---|
| 167 | search path, all the used packages that reside in this same root |
|---|
| 168 | are visible (which explains that the <b>cmt show uses</b> |
|---|
| 169 | works). |
|---|
| 170 | |
|---|
| 171 | <li> but, when you ask for <b>cmt show packages</b>, this is |
|---|
| 172 | performed outside of any context (mainly because it can be done |
|---|
| 173 | from any location). Then only packages reachable in the search |
|---|
| 174 | path (ie. <b>CMTPATH</b> or in the <b>CMT\path</b> registry set) |
|---|
| 175 | are visible. |
|---|
| 176 | </ul> |
|---|
| 177 | |
|---|
| 178 | <p>Then in your precise case, you can do one of the following : |
|---|
| 179 | |
|---|
| 180 | <ol> |
|---|
| 181 | |
|---|
| 182 | <li>If you strictly have (as you explained) |
|---|
| 183 | |
|---|
| 184 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 185 | <b> |
|---|
| 186 | <pre> |
|---|
| 187 | xxx/Velo/VeloEvent/v3/.. |
|---|
| 188 | /VSicbCnv/v3/.. |
|---|
| 189 | </pre> |
|---|
| 190 | </b> |
|---|
| 191 | </font> |
|---|
| 192 | |
|---|
| 193 | <p>then the <b>CMTPATH</b> should include <b>xxx/Velo</b> in |
|---|
| 194 | order to make both <b>VeloEvent</b> and <b>VSicbCnv</b> |
|---|
| 195 | visible. |
|---|
| 196 | |
|---|
| 197 | <li> If <b>Velo</b> is itself a package : |
|---|
| 198 | |
|---|
| 199 | <font face="courier new, courier" COLOR="#770077"> |
|---|
| 200 | <b> |
|---|
| 201 | <pre> |
|---|
| 202 | xxx/Velo/VeloEvent/v3/.. |
|---|
| 203 | /VSicbCnv/v3/.. |
|---|
| 204 | /v1/... |
|---|
| 205 | </pre> |
|---|
| 206 | </b> |
|---|
| 207 | </font> |
|---|
| 208 | |
|---|
| 209 | <p>then the <b>CMTPATH</b> may simply include <b>xxx</b> |
|---|
| 210 | (although <b>xxx/Velo</b> is possible too but in this case, it |
|---|
| 211 | is redundent). This makes all packages below <b>Velo</b> |
|---|
| 212 | visible. |
|---|
| 213 | |
|---|
| 214 | </ol> |
|---|
| 215 | |
|---|
| 216 | <p>The difference between 1) and 2) is really the fact that in 2) |
|---|
| 217 | <b>Velo</b> IS a true <b>CMT</b> package, whereas, in 1) |
|---|
| 218 | <b>Velo</b> is just a prefix. |
|---|
| 219 | |
|---|
| 220 | <p>For simple efficiency reasons, <b>CMT</b> does not scan any |
|---|
| 221 | arbitrary directory tree depth when it looks for all packages, but |
|---|
| 222 | only the tree of true <b>CMT</b> packages + one level below each |
|---|
| 223 | package. |
|---|
| 224 | |
|---|
| 225 | <hr> |
|---|
| 226 | <address><a href="mailto:arnault@lal.in2p3.fr">Christian Arnault</a></address> |
|---|
| 227 | <!-- Created: Thu Oct 26 09:09:01 MET DST 2000 --> |
|---|
| 228 | <!-- hhmts start --> |
|---|
| 229 | Last modified: Thu Nov 9 15:16:35 MET 2000 |
|---|
| 230 | <!-- hhmts end --> |
|---|
| 231 | </body> |
|---|
| 232 | </html> |
|---|
