[Scons-dev] Anyone any hints…

[Dirk Bächle]

Hi Russel,

On 05.09.2015 10:48, Russel Winder wrote:
> On Tue, 2015-09-01 at 23:14 +0200, Dirk Bächle wrote:
>> On 01.09.2015 22:37, Dirk Bächle wrote:
>> > Hi,
>> >
>> > just checked this...and it looks like I forgot to add the "suffix"
>> > attribute to the "__getattr__" for backward compatibility. I'll
>> > prepare a pull request, which should only take a few minutes.
>> >
>>
>> Done, it's PR #245...
>
> I just updated my installation and the problem seems to have gone away
> :-)
>

yay, that's good to hear! Thanks for the feedback.

> Sadly, in a sense, Haskell, like D and Chapel, emphasize a preference
> for whole program compilation, i.e. one command, compile all source at
> the same time. So I have more or less stopped using SCons tools for
> these three languages, I just use Command. And then a whole slew of
> SideEffect and Clean – which is annoying.
>
> Is there a reason why SideEffect and Clean are totally separate? (So
> that if you have a side effect you have to manually clean it.)
>

Perhaps someone had orthogonality in mind, while designing the "rules" for these built-in methods? There might be tools out there 
that collect and transport important information in the "SideEffect" file between the single build actions. So always cleaning the 
SideEffect would be the wrong thing to do.

If you regard the task of specifying both, Clean and SideEffect, as too cumbersome for your use cases, maybe a little pseudoBuilder 
would fit? It might well call Command() internally, as you're doing now. You're basically wrapping your current approach in a single 
Builder and can call it anytime.

Hope this gives you some new ideas... ;)

Best regards,

Dirk