Re: gEDA-user: needing --disable-update-mime-database

[Dan McMahill <dan@xxxxxxxxxxxx>]

Peter TB Brett wrote:
> On Friday 16 May 2008 12:12, Dan McMahill wrote:
>> Ales Hvezda wrote:
>>> Hi,
>>>
>>>> Hi, I noticed the new geda and pcb need the configure flags:
>>>> --disable-update-desktop-database --disable-update-mime-database
>>>>
>>>> since I do not have these binaries.
>>>>
>>>> For geda I have to edit the Makefile to add them which is a little odd I
>>>> think .  I just thought it would be nicer if they were autodetected
>>>> instead, just my suggestion.
>>> 	Yeah, I would like that too.  Is this sort of auto detect
>>> feasible for the desktop integration, Peter?  A possible alternative
>>> is to add (disable) them by default.  I seem to recall having a similar
>>> conversation a while ago, but I don't remember what came of it.  Thanks,
>> For what its worth, I very much dislike the proposed sort of autodetect
>> behavior.  In the past I've seen this lead to mysterious behavior like 2
>> builds configured the same having different features and a difference
>> list of dependencies.
>>
>> I would not be opposed to having them disabled by default though.
> 
> I would strongly oppose having them disabled by default, because it means that 
> (a) people won't test them and (b) people will complain about missing 
> features that aren't.

yeah, good point.

> Furthermore, not being able to run update-desktop-database and 
> update-mime-database isn't actually a big deal -- if you don't have those 
> tools, it means that they would have been noops anyway (i.e. you don't have 
> the freedesktop databases which they manipulate).

Here is my fear with the autodetection.  Suppose you have the desktop 
integration tools, but for whatever reason configure didn't find them. 
Maybe you didn't have your path set right.  Or maybe someone is creating 
a 3rd party package and forgot to list the integration stuff as a 
dependency.  Now it configures and builds and installs with no warning 
(unless you read through the very long build log) that it didn't find 
something and hence won't run the database update.  Thats why I'd rather 
have the user explicitly acknowledge that something wasn't found.  I 
think this is probably a bigger deal with libraries where it is probably 
more common for someone to have a library installed in a different spot 
and need to give LDFLAGS= arguments.  This too is a case where I'd 
rather see configure say "hey, I didn't find libfoo.  If this is ok, 
build with --disable-libfoo but you will miss out on feature xyz" 
instead of silently dropping feature xyz.

-Dan


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user