Bootstrapping from a non-checked-out distribution is risky

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Bootstrapping from a non-checked-out distribution is risky

Dmitri A. Sergatskov
It looks like after

Change #3670
Category     None
Changed by     John W. Eaton <[hidden email]>
Changed at     Fri 20 Oct 2017 16:11:58
Repository     http://hg.savannah.gnu.org/hgweb/octave
Project     octave
Branch     default
Revision     312c00dd723af2a467045b2cbbc920e74d31d083

bootstrap is not working.

I get message


./bootstrap: Bootstrapping from a non-checked-out distribution is risky.

buildbots are not happy either:


Dmitri.
--

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

John W. Eaton
Administrator
On 10/21/2017 06:48 PM, Dmitri A. Sergatskov wrote:

> It looks like after
>
> Change #3670
> Category     None
> Changed by     John W. Eaton <[hidden email] <mailto:[hidden email]>>
> Changed at     Fri 20 Oct 2017 16:11:58
> Repository http://hg.savannah.gnu.org/hgweb/octave
> Project     octave
> Branch     default
> Revision     312c00dd723af2a467045b2cbbc920e74d31d083
>
> bootstrap is not working.
>
> I get message
>
>
> ./bootstrap: Bootstrapping from a non-checked-out distribution is risky.
>
> buildbots are not happy either:
>
> http://buildbot.octave.org:8010/waterfall

Oops.  This change should fix it:

   http://hg.savannah.gnu.org/hgweb/octave/rev/ef59a196c155

jwe

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

PhilipNienhuis
John W. Eaton wrote
> On 10/21/2017 06:48 PM, Dmitri A. Sergatskov wrote:
>> It looks like after
>>
>> Change #3670
>> Category     None
>> Changed by     John W. Eaton &lt;

> jwe@

>  &lt;mailto:

> jwe@

> &gt;>
>> Changed at     Fri 20 Oct 2017 16:11:58
>> Repository http://hg.savannah.gnu.org/hgweb/octave
>> Project     octave
>> Branch     default
>> Revision     312c00dd723af2a467045b2cbbc920e74d31d083
>>
>> bootstrap is not working.
>>
>> I get message
>>
>>
>> ./bootstrap: Bootstrapping from a non-checked-out distribution is risky.
>>
>> buildbots are not happy either:
>>
>> http://buildbot.octave.org:8010/waterfall
>
> Oops.  This change should fix it:
>
>    http://hg.savannah.gnu.org/hgweb/octave/rev/ef59a196c155
>
> jwe

I think one more fix is required.
When doing "make all dist" I now get:

cp: cannot stat '../dev_p/README': No such file or directory
Makefile:26632: recipe for target 'distdir' failed
make[1]: *** [distdir] Error 1
make[1]: Leaving directory '/home/philip/devel/octdev/oct430+_p'
Makefile:26732: recipe for target 'dist' failed


Philip



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

siko1056
PhilipNienhuis wrote

> John W. Eaton wrote
>> On 10/21/2017 06:48 PM, Dmitri A. Sergatskov wrote:
>>> It looks like after
>>>
>>> Change #3670
>>> Category     None
>>> Changed by     John W. Eaton &lt;
>
>> jwe@
>
>>  &lt;mailto:
>
>> jwe@
>
>> &gt;>
>>> Changed at     Fri 20 Oct 2017 16:11:58
>>> Repository http://hg.savannah.gnu.org/hgweb/octave
>>> Project     octave
>>> Branch     default
>>> Revision     312c00dd723af2a467045b2cbbc920e74d31d083
>>>
>>> bootstrap is not working.
>>>
>>> I get message
>>>
>>>
>>> ./bootstrap: Bootstrapping from a non-checked-out distribution is risky.
>>>
>>> buildbots are not happy either:
>>>
>>> http://buildbot.octave.org:8010/waterfall
>>
>> Oops.  This change should fix it:
>>
>>    http://hg.savannah.gnu.org/hgweb/octave/rev/ef59a196c155
>>
>> jwe
>
> I think one more fix is required.
> When doing "make all dist" I now get:
>
> cp: cannot stat '../dev_p/README': No such file or directory
> Makefile:26632: recipe for target 'distdir' failed
> make[1]: *** [distdir] Error 1
> make[1]: Leaving directory '/home/philip/devel/octdev/oct430+_p'
> Makefile:26732: recipe for target 'dist' failed

Dear Philip,

On my machine, I observed the same problem.  It is a documented "feature"
(bad hard-coded design in my opinion), that each automake generated package
has to have literally: "INSTALL, NEWS, README, AUTHORS, and ChangeLog, plus
one of COPYING.LIB, COPYING.LESSER or COPYING" [1].  There is nothing you
can do, no matter how hard you try, despite of postprocessing the output of
Makefile.in or generating a literal README file [2] from README.md (I thinkt
the guys in [2] did not try `make dist`.

My suggestion is:

1. Leave etc/HACKING.md as is (it is no requirement of automake to have
one).
2. Rename README.md (with its Markdown syntax) back to README.
3. Create for doxygen a symbolic link, called "doc/doxygen/README.md" to
README to fool doxygen about the missing extension.

This changed only happened because of my attempt to create a more useful
doxygen documentation [3,4].

Kai

[1]: https://www.gnu.org/software/automake/manual/html_node/Gnits.html
[2]:
https://stackoverflow.com/questions/15013672/use-autotools-with-readme-md
[3]:
http://octave.1599824.n4.nabble.com/Making-more-of-Doxygen-td4685268.html
[4]: https://octave.space/doxygen/dev/



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

siko1056
siko1056 wrote
> [...] It is a documented "feature"
> (bad hard-coded design in my opinion), that each automake generated
> package
> has to have literally: "INSTALL, NEWS, README, AUTHORS, and ChangeLog,
> plus
> one of COPYING.LIB, COPYING.LESSER or COPYING" [1].  There is nothing you
> can do, no matter how hard you try [...]

A small addition to avoid repeating my failed attempt to overcome this
problem:

AM_INIT_AUTOMAKE([foreign

did not work either.

Kai.




--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

John W. Eaton
Administrator
In reply to this post by siko1056
On 10/23/2017 07:12 AM, siko1056 wrote:

> On my machine, I observed the same problem.  It is a documented "feature"
> (bad hard-coded design in my opinion),

OTOH, consistency can be a good thing when looking at a new project.

> 2. Rename README.md (with its Markdown syntax) back to README.

> 3. Create for doxygen a symbolic link, called "doc/doxygen/README.md" to
> README to fool doxygen about the missing extension.

Do you mean to have a symbolic link in the hg archive?  How does that
work for someone who checks out the sources on a Windows system?

Does the name of a markdown file have to end in ".md"?  Maybe that's
also a bad design decision.  ;-)

jwe

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

siko1056
John W. Eaton wrote
>> On my machine, I observed the same problem.  It is a documented "feature"
>> (bad hard-coded design in my opinion),
>
> OTOH, consistency can be a good thing when looking at a new project.

I fully agree with you.


John W. Eaton wrote
>> 2. Rename README.md (with its Markdown syntax) back to README.

This should happen soon, as I am afraid to have broken many peoples workflow
by this :(


John W. Eaton wrote
>> 3. Create for doxygen a symbolic link, called "doc/doxygen/README.md" to
>> README to fool doxygen about the missing extension.
>
> Do you mean to have a symbolic link in the hg archive?  How does that
> work for someone who checks out the sources on a Windows system?

Ah no, I would never store a symbolic link inside a repository.  My first
idea was to generate this link on building the Doxygen documentation, but
maybe it is superior, to create a renamed copy of README.  The last thing I
want was to create more unmaintainded and duplicate documentation and I am
very convinced, that the content of README and etc/HACKING is a good
starting point when digging into the development of Octave.


John W. Eaton wrote
> Does the name of a markdown file have to end in ".md"?  Maybe that's
> also a bad design decision.  ;-)

I don't like the Windows 8.3 convention either. On the other hand, the idea
of an hint (suffix or MIME type) for interpretation makes sense for me.
".markdown" might be more intuitive, also ".text" or ".plaintext" instead of
".txt".  But I am afraid, that I will not change this, so I am sure not
change the conventions autotools myself ^^

Kai



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

Reply | Threaded
Open this post in threaded view
|

Re: Bootstrapping from a non-checked-out distribution is risky

siko1056
Hopefully by cset http://hg.savannah.gnu.org/hgweb/octave/rev/ab6a631abe52
the situation is reverted back to normal.

Please tell me, if not.

Kai.



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html