GSoC Proposal : Automating the release process of sympy
92 views
Skip to first unread message
Ramana Venkata
Mar 7, 2013, 11:25:41 AM3/7/13
to sy...@googlegroups.com
Recently in one of the thread I saw somebody mentioning 'automating the release process of sympy' to be a GSoC idea for 2013 and also vaguely discussed with Aaron on IRC channel. I want to work on this idea. I have submitted a pull request and waiting for the review.
I have read the discussion in the following thread https://groups.google.com/forum/#!topic/sympy/UfNhyFv-oMg/discussion . i think the goals of the automating process are broadly the following:
-> Run all the tests mentioned in New Release page
-> Change the version numbers and create tar balls of the source
-> Upload the tar balls to necessary sites
-> Upload new documentation for the new release at http://docs.sympy.org
-> Change year in necessary places at the start of every year and other miscellaneous things
I have also looked at numpy-vendor which Ondrej has suggested. I have been familiarising myself with Fabric and Vagrant softwares currently for this idea.
I am just writing this thread to see the general acceptance of this idea as a part of GSoC in our community. I haven't presently planned on how to implement this but if this idea gets accepted I will write an in-depth proposal.
I have read the discussion in the following thread https://groups.google.com/forum/#!topic/sympy/UfNhyFv-oMg/discussion . i think the goals of the automating process are broadly the following:
-> Run all the tests mentioned in New Release page
-> Change the version numbers and create tar balls of the source
-> Upload the tar balls to necessary sites
-> Upload new documentation for the new release at http://docs.sympy.org
-> Change year in necessary places at the start of every year and other miscellaneous things
I have also looked at numpy-vendor which Ondrej has suggested. I have been familiarising myself with Fabric and Vagrant softwares currently for this idea.
I am just writing this thread to see the general acceptance of this idea as a part of GSoC in our community. I haven't presently planned on how to implement this but if this idea gets accepted I will write an in-depth proposal.
Aaron Meurer
Mar 7, 2013, 10:23:09 PM3/7/13
to sy...@googlegroups.com
I can't speak for general acceptance. I certainly think myself that it
is worthy, though.
One thing that might be an issue is that the project of just creating
a good release proces is not enough to fill an entire GSoC project.
So you should consider adding some to it. My suggestion is to improve
SymPy-Bot, which despite Travis, is still useful in my opinion.
Recently I have set up an old Linux laptop to run SymPy-Bot
automatically. But "automatically" actually just means that I have
set it to run ./sympy-bot review 1850 1851 1852 ... 1900 --profile
all-tests-no-pypy (see my profile at
https://github.com/asmeurer/dotfiles/blob/dell/.sympy/sympy-bot.conf).
This runs the bot on each request, and if it manages to get to a pull
request before it actually exists, it automatically sits there and
waits until it does, checking every so often.
This much is already implemented, but it would be great to make it
smarter. Stefan used to run a bot using some hackish script
(https://gist.github.com/Krastanov/2985162 I believe) that checked for
commits that weren't tested yet. My idea of how it should work is
outlined at https://github.com/sympy/sympy-bot/issues/63. It was also
discussed on the mailing list a lot (search for around this time last
year). Basically, I think the reviews site should keep track of what
reviews are done, and you should be able to put sympy-bot in an
automated "work" mode, which would poll the reviews site for a new
pull request to review. These would be prioritized based on various
factors, like if it's been tested yet on the available platforms, or
if it's very active, and so on.
Also, currently my laptop is just sitting in my closet, and I check on
it every once in a while. But I would like to be able to ssh into it
from my main laptop and manage everything. In addition to some tasks
that could probably be done automatically, like occasionally doing a
"git pull" in the sympy-bot repo, occasionally doing a "git pull;
./bin/use2to3" in the sympy repo (since it copies that over, and to
make things faster for testing in Python 3), there are also things
that need to be done manually, like making sure that it doesn't die.
So it would be nice to have some basic infrastructure on this, as well
as some documentation on how to do it (I am not very good with setting
up Linux servers, and I imagine others aren't as well).
I encourage you to read through all the open issues for sympy-bot
(https://github.com/sympy/sympy-bot/issues?state=open), and also
search for a similar proposal and its discussion from last year.
Regarding your ideas so far, I take it you've read my mailing list
post linked to on issue. I think you have oversimplified what needs to
be done. Some stuff you missed:
- Getting the list of AUTHORS (including making sure that the AUTHORS
and .mailmap files are up-to-date).
- Writing the release notes. There's not much we can do to automate
this, but there is some. For example, literally all changes these days
come in pull requests, so to find what has changed in a release, it is
enough to look through all the pull requests that were merged in that
release. A tool that automatically listed these in a nice way would
make writing the release notes much easier.
- There are several sites that we need to update. We can probably
forgo updating any site not owned by us (of the ones listed at the
bottom of https://github.com/sympy/sympy/wiki/new-release), but there
are several that are, such as the homepage, sympy-live, sympy-gamma,
and the blog.
- It would be nice if we could somehow keep the "dev" docs up-to-date
automatically. Ondrej probably still has a server somewhere that can
do this (he must, because something is updating Planet SymPy). It
would also be cool if we could somehow add a "dev" version to SymPy
Live and SymPy Gamma. Of course, if we start releasing once a week,
this will be completely unnecessary.
- There are dozens of little things, some of which are mandatory, and
some of which would just be nice, that you can implement. I can help
you work through an exact release process, and you can see just how
much work it really is (though the wiki page should already give you
an idea). For example, it would be nice if the coverage_doctest
script checked which docstrings are imported into Sphinx, so that we
not only have good documentation coverage in the code, but in the
online docs. This would probably take no more than a day to write,
but there are dozens of little things like this that relate to the
whole idea of automating our workflow.
I hope that gives you some ideas. I would wait to see how others feel
about this idea. You might want to come up with a secondary proposal
in case we decide we don't want this one. Your work so far looks
promising, so I would hate to see your GSoC chances destroyed just
because you picked a project that we decided we didn't want.
Finally, if you want to pursue this idea, I would suggest making
another patch, which is more inline with the idea presented here, so
that we can see that you are capable of this work too (that isn't to
say we don't like your erfc patch, but more is always better).
Perhaps a fix to sympy-bot, for example.
Aaron Meurer
> --
> You received this message because you are subscribed to the Google Groups
> "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sympy+un...@googlegroups.com.
> To post to this group, send email to sy...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sympy?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
is worthy, though.
One thing that might be an issue is that the project of just creating
a good release proces is not enough to fill an entire GSoC project.
So you should consider adding some to it. My suggestion is to improve
SymPy-Bot, which despite Travis, is still useful in my opinion.
Recently I have set up an old Linux laptop to run SymPy-Bot
automatically. But "automatically" actually just means that I have
set it to run ./sympy-bot review 1850 1851 1852 ... 1900 --profile
all-tests-no-pypy (see my profile at
https://github.com/asmeurer/dotfiles/blob/dell/.sympy/sympy-bot.conf).
This runs the bot on each request, and if it manages to get to a pull
request before it actually exists, it automatically sits there and
waits until it does, checking every so often.
This much is already implemented, but it would be great to make it
smarter. Stefan used to run a bot using some hackish script
(https://gist.github.com/Krastanov/2985162 I believe) that checked for
commits that weren't tested yet. My idea of how it should work is
outlined at https://github.com/sympy/sympy-bot/issues/63. It was also
discussed on the mailing list a lot (search for around this time last
year). Basically, I think the reviews site should keep track of what
reviews are done, and you should be able to put sympy-bot in an
automated "work" mode, which would poll the reviews site for a new
pull request to review. These would be prioritized based on various
factors, like if it's been tested yet on the available platforms, or
if it's very active, and so on.
Also, currently my laptop is just sitting in my closet, and I check on
it every once in a while. But I would like to be able to ssh into it
from my main laptop and manage everything. In addition to some tasks
that could probably be done automatically, like occasionally doing a
"git pull" in the sympy-bot repo, occasionally doing a "git pull;
./bin/use2to3" in the sympy repo (since it copies that over, and to
make things faster for testing in Python 3), there are also things
that need to be done manually, like making sure that it doesn't die.
So it would be nice to have some basic infrastructure on this, as well
as some documentation on how to do it (I am not very good with setting
up Linux servers, and I imagine others aren't as well).
I encourage you to read through all the open issues for sympy-bot
(https://github.com/sympy/sympy-bot/issues?state=open), and also
search for a similar proposal and its discussion from last year.
Regarding your ideas so far, I take it you've read my mailing list
post linked to on issue. I think you have oversimplified what needs to
be done. Some stuff you missed:
- Getting the list of AUTHORS (including making sure that the AUTHORS
and .mailmap files are up-to-date).
- Writing the release notes. There's not much we can do to automate
this, but there is some. For example, literally all changes these days
come in pull requests, so to find what has changed in a release, it is
enough to look through all the pull requests that were merged in that
release. A tool that automatically listed these in a nice way would
make writing the release notes much easier.
- There are several sites that we need to update. We can probably
forgo updating any site not owned by us (of the ones listed at the
bottom of https://github.com/sympy/sympy/wiki/new-release), but there
are several that are, such as the homepage, sympy-live, sympy-gamma,
and the blog.
- It would be nice if we could somehow keep the "dev" docs up-to-date
automatically. Ondrej probably still has a server somewhere that can
do this (he must, because something is updating Planet SymPy). It
would also be cool if we could somehow add a "dev" version to SymPy
Live and SymPy Gamma. Of course, if we start releasing once a week,
this will be completely unnecessary.
- There are dozens of little things, some of which are mandatory, and
some of which would just be nice, that you can implement. I can help
you work through an exact release process, and you can see just how
much work it really is (though the wiki page should already give you
an idea). For example, it would be nice if the coverage_doctest
script checked which docstrings are imported into Sphinx, so that we
not only have good documentation coverage in the code, but in the
online docs. This would probably take no more than a day to write,
but there are dozens of little things like this that relate to the
whole idea of automating our workflow.
I hope that gives you some ideas. I would wait to see how others feel
about this idea. You might want to come up with a secondary proposal
in case we decide we don't want this one. Your work so far looks
promising, so I would hate to see your GSoC chances destroyed just
because you picked a project that we decided we didn't want.
Finally, if you want to pursue this idea, I would suggest making
another patch, which is more inline with the idea presented here, so
that we can see that you are capable of this work too (that isn't to
say we don't like your erfc patch, but more is always better).
Perhaps a fix to sympy-bot, for example.
Aaron Meurer
> You received this message because you are subscribed to the Google Groups
> "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to sympy+un...@googlegroups.com.
> To post to this group, send email to sy...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sympy?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
Ramana Venkata
Mar 9, 2013, 5:53:21 AM3/9/13
to sy...@googlegroups.com
Thanks Aaron for your valuable suggestions. I will do some work on Sympy-bot along side. Improving Sympy bot Is this the proposal you were referring to??
I am presently having hard time with sympy-bot. When I trying to run ./sympy-bot list (I had setup token in sympy-bot.confand the API token as well). I am facing this issue https://github.com/sympy/sympy-bot/issues/147. I have read about the rate limiting and basic authentication and other stuff. But still I couldn't understand how to make my sympy-bot requests authenticated.
I want to understand the overall sympy-bot code but it's not going well. I feel that sympy-bot should have a little more documentation.
I am presently having hard time with sympy-bot. When I trying to run ./sympy-bot list (I had setup token in sympy-bot.confand the API token as well). I am facing this issue https://github.com/sympy/sympy-bot/issues/147. I have read about the rate limiting and basic authentication and other stuff. But still I couldn't understand how to make my sympy-bot requests authenticated.
I want to understand the overall sympy-bot code but it's not going well. I feel that sympy-bot should have a little more documentation.
Aaron Meurer
Mar 9, 2013, 1:27:48 PM3/9/13
to sy...@googlegroups.com
On Saturday, March 9, 2013, Ramana Venkata wrote:
Thanks Aaron for your valuable suggestions. I will do some work on Sympy-bot along side. Improving Sympy bot Is this the proposal you were referring to??
Yes. I think there was even more discussion than that, though. Also earch the wiki and irc logs.
I am presently having hard time with sympy-bot. When I trying to run ./sympy-bot list (I had setup token in sympy-bot.confand the API token as well). I am facing this issue https://github.com/sympy/sympy-bot/issues/147. I have read about the rate limiting and basic authentication and other stuff. But still I couldn't understand how to make my sympy-bot requests authenticated.
I want to understand the overall sympy-bot code but it's not going well. I feel that sympy-bot should have a little more documentation.
Unfortunately, SymPy bot started out as kind of a thrown together hack, and it still retains some of that feel.
Aaron Meurer
Ramana Venkata
Mar 9, 2013, 1:40:22 PM3/9/13
to sy...@googlegroups.com
Yeah :) I have seen the thread Ondrej has written when he started sympy-bot. I have read through the issues list in sympy-bot but I couldn't figure out what are some easy to fix/implement issues available on the list. So can you list out a few??
How to fix API limit exceeded issue while trying to run ./sympy-bot list? I have created sympy-bot.conf and oauth token. I have read through http://developer.github.com/v3/#rate-limiting but I couldn't figure out anything.--
You received this message because you are subscribed to a topic in the Google Groups "sympy" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sympy/v7bw8smG444/unsubscribe?hl=en.
To unsubscribe from this group and all its topics, send an email to sympy+un...@googlegroups.com.
To post to this group, send email to sy...@googlegroups.com.
Visit this group at http://groups.google.com/group/sympy?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
--
Venkata Ramana K.
Sophomore,Aaron Meurer
Mar 9, 2013, 2:11:45 PM3/9/13
to sy...@googlegroups.com
On Mar 9, 2013, at 11:40 AM, Ramana Venkata <idlike...@gmail.com> wrote:
Yeah :) I have seen the thread Ondrej has written when he started sympy-bot. I have read through the issues list in sympy-bot but I couldn't figure out what are some easy to fix/implement issues available on the list. So can you list out a few??
I can try to tag them more specifically later, but for now, the ones with the code-in labels should be easier.
How to fix API limit exceeded issue while trying to run ./sympy-bot list? I have created sympy-bot.conf and oauth token. I have read through http://developer.github.com/v3/#rate-limiting but I couldn't figure out anything.
I think you just need to change all _query calls to pass in the password or token when it's known. You might also have to make sure that list actually gets the token.
Aaron Meurer
Message has been deleted
Ramana Venkata
Mar 13, 2013, 10:56:21 AM3/13/13
to sy...@googlegroups.com
Sorry I accidentally deleted my previous post. I am reposting the content
The issues list in sympy-bot list contains lists which are either
related to sympy-bot code or with the review.sympy.org site's
functionality and some related to both in a way. @aaron Can you create
to labels for both of them and categorize them accordingly??
On Wed, Mar 13, 2013 at 8:23 PM, Ramana Venkata <idlike...@gmail.com> wrote:
The issues list in sympy-bot list contains lists which are either related to sympy-bot code or with the review.sympy.org site's functionality and some related to both in a way. @aaron Can you create to labels for both of them and categorize them accordingly??
To unsubscribe from this group and stop receiving emails from it, send an email to sympy+unsubscribe@googlegroups.com.
Ramana Venkata
Mar 14, 2013, 10:20:26 PM3/14/13
to sy...@googlegroups.com
@aaron I have sorted out the issues into Sympy-Bot and Reviews.sympy.org
Bot:
144 142 139 121 120 108 95 94 89 80 73 63 62 56 21 2
Review.sympy.org:
138 132 120 119 117 103 102 90 59 57 55 51 50 49 48 47 46 45 42 35 16
Are the issues 57, 48 still valid?
Bot:
144 142 139 121 120 108 95 94 89 80 73 63 62 56 21 2
Review.sympy.org:
138 132 120 119 117 103 102 90 59 57 55 51 50 49 48 47 46 45 42 35 16
Are the issues 57, 48 still valid?
Reply all
Reply to author
Forward
0 new messages