[Scons-users] Need help debugging my build

[William Blevins]

As a side note, this appears to be because SCons is following the symlinks
rather than interpreting the symlinks as targets themselves; thus, SCons
thinks that liba.so, liba.so.0, and libtest.so.0.1.0 are all the same
target.

V/R,
William

On Thu, Jun 11, 2015 at 10:17 AM, William Blevins <wblevins001 at gmail.com>
wrote:

> Dirk,
>
> I ran a simple test:
>
> SConstruct: SharedLibrary( 'a', 'a.cpp', SHLIBVERSION = '0.1.0' )
> a.cpp: int main() { return 0; }
>
> Under the following systems:
> SCons 2.3.0, CentOS 6, python 2.6
> SCons 2.3.4, CentOS 7, python 2.7
>
> I got the same results on both, so this isn't a new issue.
>
> On compile attempts 2 thru N, the link target is re-executed, and I get
> the error: scons: *** [liba.so.0.1.0] File exists
>
> V/R,
> William
>
> On Thu, Jun 11, 2015 at 8:31 AM, Dirk Bächle <tshortik at gmx.de> wrote:
>
>> Hi Gottfried (Moin!),
>>
>> On 04.06.2015 16:31, Ganssauge, Gottfried wrote:
>>
>>> My build – which is somewhat large – normally runs fine using
>>> scons-local-2.3.4 with python2.7.6 on Ubuntu Linux-14.04
>>> But when I’m starting the same build twice in a row I’d expect it to
>>> rebuild nothing at all the second time through.
>>> The actual targets which I request aren’t rebuilt in fact but some
>>> intermediate targets.
>>>
>>
>> what happens when you rebuild the same target a third time? Is the output
>> as you'd expect in run #3?
>> Then, you can try to disable the implicit command dependency mechanism
>> with
>>
>>   env['IMPLICIT_COMMAND_DEPENDENCIES'] = False
>>
>> and see if it helps.
>> The underlying problem is, that SCons stores the list of dependencies
>> (and its checksums) as ordered list (=vector, not dict). In your use case
>> the number of children and their checksums stay the same...but their order
>> changes from one run to the other. This may happen especially for files
>> that get created during a build, because they get found earlier or later
>> then while scanning folders for dependencies...depending on whether they
>> already exist, how search paths are set, and how the implicit dependency
>> graph is traversed.
>>
>> Best regards,
>>
>> Dirk
>>
>>
>> _______________________________________________
>> Scons-users mailing list
>> Scons-users at scons.org
>> https://pairlist4.pair.net/mailman/listinfo/scons-users
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist4.pair.net/pipermail/scons-users/attachments/20150611/bbc7bdb0/attachment.html>