[Scilab-Dev] slint() : remarks and suggestions

[Samuel Gougeon]

Hello,

Here are some additional remarks about the "conf" configuration file, 
and the file of results:

Le 19/04/2016 13:09, Samuel Gougeon a écrit :
> .../...
>
>       * Presently, slint() does not allow
>           o to *provide a **Scilab version against which the Scilab
>             code must be checked*.
>           o to provide a subset of rules (or categories of rules) that
>             must be checked, instead of checking all defined rules.
>
.

  * Apparently, the so-called "configuration" file is somewhat used to
    that, but it's rather unhandy: to set a set of rules, we have to
    edit the conf file, to turn enable="true"into enable="false"for each
    rule that we want to cancel (or the opposite), to save
    modifications, and then call slint() with the file as "conf"
    parameter... instead of directly passing the vector of rules
    indices, or a vector of text ids (case-insensitive, please) such as
    ["uselessarg" "FunctionName" "redefinition"], or a vector of ids of
    rules subsets. Rather cumbersome.

  * The format of the file of results is not documented. It is an XML
    file. Why has this encoding been chosen? XML is for parsers rather
    than for humans -- whereas a CSV is directly readable --, but here
    for which parser? Nothing is told about how to
    process/update/upgrade input files using results. Is there anything
    planned, for instance with Scinotes that could use input files and
    XMLed results to edit files and highlight parts to be updated, or
    even to automate the upgrade? If nothing is presently planned in
    such a way, imo encoding results in CSV would be better.

Best regards
Samuel Gougeon

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/dev/attachments/20160425/0a2e5874/attachment.htm>