[Scons-dev] Problem using doc toolchain?

Bill Deegan bill at baddogconsulting.com
Fri Jul 31 16:40:33 EDT 2015


O.k. Good.
It looks like the source of the bogus XML's was using lxml and not using
libxml2/libxslt.
Perhaps we should remove that as an option?
(even if only until someone is able to figure out why it's generating bad
output?)

Here's my blog post on building this all from scratch on ubuntu.

http://baddogconsulting.com/building-python-with-libxml2-and-libxslt-on-ubuntu-1404-from-source.html

(Need for scratch was driven by the fact that debian and there for ubuntu
strip the windows installer stubs from their python distributions)
-Bill

On Fri, Jul 31, 2015 at 11:44 AM, Gary Oberbrunner <garyo at oberbrunner.com>
wrote:

> All those look OK to me, Bill -- hmm, good thing I don't use swear words
> in my scons source dir names!  I wasn't expecting to see those in the doc!
> :-)
> There's no tool to check; you have to review by hand.  Usually the diffs
> are short like this.
>
> On Fri, Jul 31, 2015 at 1:30 PM, Bill Deegan <bill at baddogconsulting.com>
> wrote:
>
>> Did a clean python, libxml2-2.9.2, libxslt-1.28 from source (ran into an
>> issue with libpython.so from system python causing core dump with got me
>> stuck for a while because the error you get until you dig into it just
>> indicates that expat was not built with your python.. blog posting to
>> follow on that fun).
>>
>> Once I sorted expat issue, and run
>> /home/bdbaddog/tools/python-2.7.10/bin/python bin/docs-update-generated.py
>> /home/bdbaddog/tools/python-2.7.10/bin/python bin/docs-validate.py
>> /home/bdbaddog/tools/python-2.7.10/bin/python
>> bin/docs-create-example-outputs.py
>>
>> I get only the following diffs:
>> M doc/generated/examples/caching_ex-random_1.xml
>>    http://pastebin.com/UwE75eTY
>> M doc/generated/examples/troubleshoot_explain1_3.xml
>>   http://pastebin.com/3n2f3e4y
>> M doc/generated/variables.gen
>>   http://pastebin.com/UnhexDVR
>> M doc/generated/variables.mod
>>    http://pastebin.com/E1nXYupB
>>
>> Are these all valid?
>> Is there an easy way to check them (aka a tool?)
>>
>> -Bill
>>
>>
>> On Fri, Jul 31, 2015 at 6:11 AM, William Blevins <wblevins001 at gmail.com>
>> wrote:
>>
>>> Dirk,
>>>
>>> I had lxslt installed but not python-lxslt.  Once that was installed it
>>> was obvious that it switched from lxml to lxml2 usage.  I still got another
>>> error.
>>> On Jul 31, 2015 3:18 AM, "Dirk Bächle" <tshortik at gmx.de> wrote:
>>>
>>>> Bill,
>>>>
>>>> On 30.07.2015 17:36, Bill Deegan wrote:
>>>>
>>>>>  From the code I've looked at if you have libxml2 & libxslt that is
>>>>> preferred, and then if not it will use lxml.
>>>>>
>>>>>
>>>> your assumption is correct, this is done because libxml2 is faster in
>>>> general.
>>>>
>>>> It seems that libxml2 and pure lxml have different behaviour regarding
>>>> "normalizing namespaces" and that's where the diff comes from. This makes
>>>> at least the validation in the SernaFree XML editor choke for the lxml
>>>> output...:(
>>>>
>>>> I'm investigating this a little further and will try to find a way
>>>> around this. I'd really like to have the (almost) same output for both XML
>>>> toolchains, such that it gets accepted by most XML editors out there.
>>>>
>>>> @William: You said that after installing an additional package the
>>>> processing got faster and correct? My guess would be that you now have a
>>>> lxml distro/package that relies on libxml2 under the hood. This makes the
>>>> error go away of course...
>>>>
>>>>
>>>> I'll keep you posted, best regards,
>>>>
>>>> Dirk
>>>>
>>>> _______________________________________________
>>>> Scons-dev mailing list
>>>> Scons-dev at scons.org
>>>> https://pairlist2.pair.net/mailman/listinfo/scons-dev
>>>>
>>>
>>> _______________________________________________
>>> Scons-dev mailing list
>>> Scons-dev at scons.org
>>> https://pairlist2.pair.net/mailman/listinfo/scons-dev
>>>
>>>
>>
>> _______________________________________________
>> Scons-dev mailing list
>> Scons-dev at scons.org
>> https://pairlist2.pair.net/mailman/listinfo/scons-dev
>>
>>
>
>
> --
> Gary
>
> _______________________________________________
> Scons-dev mailing list
> Scons-dev at scons.org
> https://pairlist2.pair.net/mailman/listinfo/scons-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist2.pair.net/pipermail/scons-dev/attachments/20150731/410f7f7d/attachment.html>


More information about the Scons-dev mailing list