[Scons-users] 回复： 回复： can I have a feature request: do a full buildprocess in the rule collect phase

[查理王]

thanks very much, I will try it.




------------------ 原始邮件 ------------------
发件人: "Bill Deegan";<bill at baddogconsulting.com>;
发送时间: 2015年9月11日(星期五) 上午9:49
收件人: "SCons users mailing list"<scons-users at scons.org>; 

主题: Re: [Scons-users]回复： can I have a feature request: do a full buildprocess in the rule collect phase



Would the SIP builder from an older mailing list message work for you?

http://scons.1086193.n5.nabble.com/Tool-for-SIP-td14310.html


-Bill


On Thu, Sep 10, 2015 at 6:48 PM, Bill Deegan <bill at baddogconsulting.com> wrote:
Charles,


For your specific usage, how many sip files and how many generated files are we talking about?


-Bill


On Thu, Sep 10, 2015 at 6:33 PM, 查理王 <75187631 at qq.com> wrote:
Bill:


Thanks, let's discuss in scons-users at scons.org.


It is absolutely the best idea to use Scanner when possible, but some tools do not have this feature and we can not modify it if it is a close source product. SIP is not the only example.


We use "scons in scons" when we do not any other way to finish it, or it is too trouble to write a scanner of it.


Charles


------------------ 原始邮件 ------------------
发件人: "Bill Deegan";<bill at baddogconsulting.com>;
发送时间: 2015年9月11日(星期五) 上午9:20
收件人: "SCons users mailing list"<scons-users at scons.org>; 

主题: Re: [Scons-users] can I have a feature request: do a full buildprocess in the rule collect phase



Charles,


Generally running SCons from SCons (as you suggest) is not a great idea.

Secondly this message is best sent to the users mailing list and not the dev mailing list. (I see you sent it to both).


Ideally the QT tool would include a scanner to process the SIP input file(s) and add the expected output file(s) to the DAG.

If it is not doing that, then the best thing to do would be to add the logic to do that.



-Bill


On Thu, Sep 10, 2015 at 4:13 PM, 查理王 <75187631 at qq.com> wrote:
hi~ all:


As I know, the whole build process of scons consist of:
1, run the SConstruct and its SConscript to collect the build rules.
2, analyse the depencies to determine which actions have to be done.
3, do the selected actions with dependency restrict.


We should not include a long time calculate in the first phase(rule collection) because to avoid unnecessary long time action is the purpose of the use of scons and other building tools. But it will trap us when we have to build the rules by the result of some special long time and complex action.


For example, the SIP provided by Qt, one .sip can generate many files (named by sip_*.cc and sip*.h). all of the generated files have to be built and linked into the output program. In practice, SIP is a long time and complex process that we need to avoid the unnecessary run of it.


The resolution of us is call another scons process by os.system in the first phase by os.system('scons -f SConstruct.SIP') and then env.Program('tester', ['main.cpp'] + glob('sip_*.c'')).


We believe that this is a common way to resolve this kind of problem, but the os.system('scons -f ...') is to urgly and if we can encapsulate the whole build process into a class, then the environment and some other variables can be shared in main and sub build process. And the user will not be confused to see multiple scons running prompt in a single scons command.


The source of /usr/bin/scons and /usr/lib/scons/SCons/Script are checked, unlike the other part of scons which is in highly objective way, these part of work is not encapsulated into class yet. If these code can be encapsulated and provided into SConstruct runtime environment, it should be an important enhancement of scons.


Thanks very much.


Charles.




_______________________________________________
 Scons-users mailing list
 Scons-users at scons.org
 https://pairlist4.pair.net/mailman/listinfo/scons-users
 







_______________________________________________
 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/20150911/0bbeab9a/attachment-0001.html>