Re: [HACKERS] PostGreSQL v6.2.1 for Linux Alpha

Поиск
Список
Период
Сортировка
От Bruce Momjian
Тема Re: [HACKERS] PostGreSQL v6.2.1 for Linux Alpha
Дата
Msg-id 199802151747.MAA25325@candle.pha.pa.us
обсуждение исходный текст
Ответ на Re: [HACKERS] PostGreSQL v6.2.1 for Linux Alpha  ("Kenji T. Hollis" <khollis@Gawain.Houston-InterWeb.COM>)
Список pgsql-hackers
> I totally understand where you're coming from, however, I wrote over 4
> months ago reporting the problem, and was kind of brushed off like it was
> a non-issue, and I reported it twice.  Same thing twice.  Finally, I raise
> hell about it and I get a response.  I'm not impatient.  I am currently
> using msql for what I need and it seems to be getting the job done.  I
> just need somethimg much more powerful.  PostGres is much more powerful.
>
> I'm trying to help, but the first message I get was from some idiot on
> this list who told me to fix the problem myself and quit whining.  I don't
> take kindly to people who write messages like those to me, especially when
> I'm trying to help resolve an issue with the coders.  I don't deserve to
> be treated with this disrespect, and I never shoot back at any question I
> get - the most I'll do is delete the message and bite my damn tongue.
>
> So, now that you understand my position, I just wanted an update, and
> wanted to help fix the problem.
>
> > PS. Kenji, where did you get that pgsql would be running on Linux/Alpha in
> > Jan '98? I have never seen any such projected date!
>
> One of the coders told me this privately.  I wish I had saved that
> message!

Ask and you shall receive.  Here is the message.  At the time of 6.2,
Alpha/Linux was broken because the inlinine of asm() would not work
because alpha/linux does not support local labels in asm statements, so
if multiple lock requests were in the same C file, you had multiple
labels with the same name, and it would not compile.

This was worked on and a new fix was made, but that did not work because
of the labels problem, so the attached fix was also broken.  The final
solution was to move the alpha/linux lock code to a separate C file, so
there was every feeling it would now work in 6.3.

Turns out there were other problems related to compiler flags, and that
has only recently been added to the source.  Good thing Kenji spoke up
so we could get all the alpha/linux people together on this.

---------------------------------------------------------------------------

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Bruce Momjian
Дата:
Сообщение: Adding README files to source
Следующее
От: Bruce Momjian
Дата:
Сообщение: Re: [HACKERS] spin locks