[Olena-patches] cleanup-2008 2538: Continue documentation.

https://svn.lrde.epita.fr/svn/oln/branches/cleanup-2008/milena Index: ChangeLog from Nicolas Ballas <ballas(a)lrde.epita.fr> Continue documentation. * doc/tutorial/design/include/properties.tex: Add properties. * doc/tutorial/design/include/imagetours.tex: Fix typos. * doc/tutorial/design/design.tex: Update. design.tex | 14 ++++++++ include/imagetours.tex | 6 +-- include/properties.tex | 83 ++++++++++++++++++++++++++++++++++++++++--------- 3 files changed, 85 insertions(+), 18 deletions(-) Index: doc/tutorial/design/include/properties.tex --- doc/tutorial/design/include/properties.tex (revision 2537) +++ doc/tutorial/design/include/properties.tex (working copy) @@ -1,6 +1,7 @@ \section{Properties} + \subsection{Notation} A property has several values. @@ -107,13 +108,12 @@ between its sites. An image on a non-localized space has this property set to $none$. An image on a localized space has this property set to $space$. -An image based on a isotropic grid has this property set to $isotropic\_gride$ +An image based on a isotropic grid has this property set to $isotropic\_grid$ An image based on a anisotropic grid has this property set to -$anisotropic\_gride$ +$anisotropic\_grid$ An image based on a aligned and orthogonal grid has its property set to $basic\_grid$ -%% FIXME: Is it true? If an image has its \verb+localization+ set to $space$, the user must define the neighborhood relationships between the sites. Otherwise, the image underlying grid defines the neighborhood relationships. @@ -139,8 +139,6 @@ +--- two_d | +--- three_d - - \end{verbatim} This property specifies the dimension of the image. @@ -159,7 +157,7 @@ \index{Property: value\_access} \begin{verbatim} -value\_access: any +value_access: any | +--- direct | @@ -170,19 +168,19 @@ Image values can either be computed on the fly by a function or stored in -memory. If an image type has a direct acces on the values in memory, it is +memory. If an image type has a direct access on the values in memory, it is possible to take a reference of them. In this case, this property is set to $direct$. In the other case, this property is set $indirect$. When an image computes its values on the fly, this property is refined to $computed$. -\subsubsection{vw\_storage} -\index{Property: vw\_storage} +\subsubsection{value\_storage} +\index{Property: value\_storage} \begin{verbatim} -vw_storage: any +value_storage: any | - +--- /ram/ + +--- /memory/ | | | +--- singleton | | @@ -192,6 +190,37 @@ +--- irrelevant \end{verbatim} +This property indicates how the values are stored in memory, when it is the +case. If the values aren't stored in the $memory$, this property is set to +$irrelevant$. If this property is set to $singleton$, only one value is store +in memory for the whole image (this the case of the \verb+flat_image+ type). +If this value is set to $one\_block$, all the image values are stored in a +linear buffer in memory. And, an image is $piecewise$, when it's value are +stored in several buffers in memory. + + +\subsubsection{value\_alignment} +\index{Property: value\_alignment} + +\begin{verbatim} +value_alignment: any + | + +--- with_grid + | + +--- not_aligned + | + +--- irrelevant +\end{verbatim} + +This property indicates if the image values are ordered the same way in memory +than when we run over them through a \verb+fwd_piter+. +When this is the case, the image values are aligned with the image underlying +grid, this property is set to $with\_grid$. +If the values are not ordered the same way in memory than the \verb+fwd_piter+ +order, this property is set to $not\_aligned$. +If the image values are not stored in memory, this property is set to +$irrelevant$. + %%%%%%%%%%%%%%%%%%%%%%%% % access % @@ -255,23 +284,38 @@ %% FIXME functions that modify globaly all the values... \begin{verbatim} - vw_io: any | - +--- /vm_some/ + +--- /some/ | | | +--- vw_read | | | +--- vw_read_write - +--- vw_none + | | + | +--- vw_global_read_write + +--- none \end{verbatim} +This property defines the value wise access of an image. +If \verb+vw_io+ is set to $vw\_none$, the associated image type does not + provide any value wise access. +If an image type provides a readable (or writable) +value access, this property is set to $vw\_read$ (or $vw\_read\_write$). +If the image provide a value wise access that allows the user to change +globally all the image values through a function $value$ to $value$, +this property is set to $vw\_global\_read\_write$. + +%% FIXME define the interface associated + +The method provided for this value access depend on the \verb+vm_set+ property. + + \subsubsection{vw\_set} \index{Property: vw\_set} \begin{verbatim} - vw_io: any + vw_set: any | +--- /some/ | | @@ -282,6 +326,15 @@ \end{verbatim} +This property precises what kind of value access is provided by an image type. +If an image type does not provide any value access (\verb+vw_io+ set to $none$) +, \verb+vw_set+ is set to $none$. +Otherwise, \verb+vw_set+ is set to $uni$ if the image provides an +access to the mathematical set of values taken by the image (through the method +\verb+value_taken+). +If \verb+vw_set+ is set to $multi$, the image provides an access to the values +taken in memory (cells) (through the method \verb+cells+). Note that a value +can be present twice in the cells set. Index: doc/tutorial/design/include/imagetours.tex --- doc/tutorial/design/include/imagetours.tex (revision 2537) +++ doc/tutorial/design/include/imagetours.tex (working copy) @@ -43,8 +43,8 @@ compose the definition domain of the image. A value is associated to each site of the image. This is the destination domain of the image. -To access to a value of an image nammed $ima$ localized at the point p, we -just use the mathematic notation: \verb+ima(p)+. +To access to a value of an image named $ima$ localized at the point p, we +just use the mathematics notation: \verb+ima(p)+. Obviously, every image types of Milena provide an access $site$ to $value$, through \verb+ima(p)+. Yet, depending on the image type, the values can be @@ -62,7 +62,7 @@ %% Explain the different template parameters T/S/F... All the image types are parametrized by different static parameters. -In this document, we will used the following namming convention for the +In this document, we will used the following naming convention for the image types parameters: \begin{itemize} \item{\verb+T+:} represents an image value type. Index: doc/tutorial/design/design.tex --- doc/tutorial/design/design.tex (revision 2537) +++ doc/tutorial/design/design.tex (working copy) @@ -35,6 +35,20 @@ stringstyle=\color{green}\ttfamily } +\lstnewenvironment{cartouche} + {\lstset{tabsize=2, + frame=single, + breaklines=true, + basicstyle=\ttfamily, + framexleftmargin=13mm, + identifierstyle=\color{black}, + keywordstyle=\color{black}, + stringstyle=\color{black}, + commentstyle=\color{black}, + backgroundcolor=\color{grey}, + xleftmargin=12mm} + } +{} %%%LISTINGS SETTINGS