Re: Knowing where a function has been used (bis) [Was: Re: Optimising Elisp code]

[Garreau, Alexandre]

On 2018-10-15 at 21:27, Emanuel Berg wrote:
> tomas wrote:
>
>> On the long term this tends lead to horrible
>> code, where the defuns reflect all the
>> history of the situations, as they arrived,
>> not the problem's structure as it is *now*.
>
> Why is this horrible code? Isn't it just
> natural that the code reflects how problems
> have been solved over time, with changes in
> technology, etc.?

No, VCSes a here for that.

> Isn't it much better to have this "reflection" in neat modules than in
> single functions spanning 3 A4 each?

If these modules have no meaning, no, because it will harden the work of
understanding, thus, for instance, of debugging, the function.  While
the same functions may have piece that will cancel or make useless
pieces of other functions, if you just “add up historically”, you may,
if you want to solve the problem as it is *now*, simplify the whole
thing.  And it won’t simply be refactoring by adding functions or
macros, it will be removing stuff.

> Besides, if some of these defuns are really
> part of history with no relevance for the
> present situation, it is much easier to delete
> them [1] than to mess with the
> original huge functions!

> [1] Better yet, delete *the call* to them and
>     to move the modules to an out-of-action
>     source file, as history might
>     repeat itself.

VCS, search tools, or otherwise devs memories, are here for that.

And even if it stayed in the source code anyway people are interested in
functions that are documented and that they use, directly or indirectly.
They might as well not see it if it’s not used.