Release Plans
20 views
Skip to first unread message
David Nadlinger
Nov 7, 2012, 1:32:38 PM11/7/12
to ldc...@googlegroups.com
Hi all,
Okay, so my original plans of making a release in September turned out
to be nothing but a pipe dream; I underestimated the amount of issues
that would turn up, while at the same time being swamped by university
work much sooner than expected.
In any case, I think it is important for us to get out a release as
soon as possible, to have a solid base for working on ease of access
to LDC, including updated packages in Debian and other distributions
that generally don't ship unreleased versions.
Now, my question: In your opinion, which of the remaining bugs should
we definitely fix before working on finalizing the release?
I have my own list, but let my instead note two things which I don't
think will make it into this release:
- Fixes for the x86_64 ABI issues – I have started work on it, but
there is still quite a lot to do, and the current code _mostly_ works.
- My rewritten variadic argument code – I would have loved to see
this in the release, but I recently discovered some new issues (e.g.
Github #172), and I'm afraid it will take quite a bit more work to get
the ode in shape.
On a related note, my plan for the future is to track DMD upstream
releases as close as possible with our own releases, so that never
have to wait too long to get their hands on new frontend versions
without having to manually build from source (of course, this would be
easier if DMD had a somewhat predictable release schedule). What do
you think?
David
Okay, so my original plans of making a release in September turned out
to be nothing but a pipe dream; I underestimated the amount of issues
that would turn up, while at the same time being swamped by university
work much sooner than expected.
In any case, I think it is important for us to get out a release as
soon as possible, to have a solid base for working on ease of access
to LDC, including updated packages in Debian and other distributions
that generally don't ship unreleased versions.
Now, my question: In your opinion, which of the remaining bugs should
we definitely fix before working on finalizing the release?
I have my own list, but let my instead note two things which I don't
think will make it into this release:
- Fixes for the x86_64 ABI issues – I have started work on it, but
there is still quite a lot to do, and the current code _mostly_ works.
- My rewritten variadic argument code – I would have loved to see
this in the release, but I recently discovered some new issues (e.g.
Github #172), and I'm afraid it will take quite a bit more work to get
the ode in shape.
On a related note, my plan for the future is to track DMD upstream
releases as close as possible with our own releases, so that never
have to wait too long to get their hands on new frontend versions
without having to manually build from source (of course, this would be
easier if DMD had a somewhat predictable release schedule). What do
you think?
David
Kai
Nov 8, 2012, 2:51:49 PM11/8/12
to ldc...@googlegroups.com
On 07.11.2012 19:32, David Nadlinger wrote:
> Hi all,
>
> Okay, so my original plans of making a release in September turned out
> to be nothing but a pipe dream; I underestimated the amount of issues
> that would turn up, while at the same time being swamped by university
> work much sooner than expected.
>
> In any case, I think it is important for us to get out a release as
> soon as possible, to have a solid base for working on ease of access
> to LDC, including updated packages in Debian and other distributions
> that generally don't ship unreleased versions.
>
> Now, my question: In your opinion, which of the remaining bugs should
> we definitely fix before working on finalizing the release?
I think issue #205 "Merge DMD 1.075 frontend and get Tango tests to
> Hi all,
>
> Okay, so my original plans of making a release in September turned out
> to be nothing but a pipe dream; I underestimated the amount of issues
> that would turn up, while at the same time being swamped by university
> work much sooner than expected.
>
> In any case, I think it is important for us to get out a release as
> soon as possible, to have a solid base for working on ease of access
> to LDC, including updated packages in Debian and other distributions
> that generally don't ship unreleased versions.
>
> Now, my question: In your opinion, which of the remaining bugs should
> we definitely fix before working on finalizing the release?
pass" should be fixed. Main reason is that working tests are required
tor quality measurement. Also we should not fall behind the current DMD
development too much. Merging more then one release at once is painful.
As soon as we have an official release I would expect a lot of other
errors. For sure this will resort the priorities of the issues to fix. I
think it is more important to document/automate the release process.
Then we can provide releases with minor updates quickly.
BTW: I created a Gentoo ebuild for LDC. If you like then I can setup
another git repository and provide an official "LDC overlay".
Kai
>
> I have my own list, but let my instead note two things which I don't
> think will make it into this release:
> there is still quite a lot to do, and the current code _mostly_ works.
> - My rewritten variadic argument code � I would have loved to see
Reply all
Reply to author
Forward
0 new messages