Subject:
|
Re: NQC beta test release policy
|
Newsgroups:
|
lugnet.robotics.rcx.nqc
|
Date:
|
Fri, 7 Jan 2000 04:03:47 GMT
|
Viewed:
|
1471 times
|
| |
|
|
In article <38749E2F.789B75D6@hotpop.com>, Leonard Stiles
<lstiles@hotpop.com> wrote:
>
> In the light of this fact I find it somewhat illogical that the NQC 2.1
> beta test is a closed source binaries-only release. Not only does it
> exclude those unable to run mac/win32 binaries, but it also makes the
> bugfix process more difficult.
>
> I would be interested to hear thoughts on this.
Illogical yes, but not without its reasons.
The exclusion of linux and source from beta isn't really a result of any
policy, but just to make things more convenient for me. Here's why...
I don't have 24/7 access to a linux box, so making a linux release usually
delays things a couple of days until I can get a chance to do the build.
Early on, beta releases came out pretty fast and were short lived so
adding even a couple of days of delay to the beta release process seemed
counter-productive.
For various reasons the source release is tied to the linux release:
* The source release requires the Makefile to be updated and verified -
this only happens on the linux build since mac/win use a CodeWarrior
project file.
* The source release takes several steps - including building default
versions of the parser files, etc. All of this is automated by some
scripts and Makefiles I have running under Linux. I'm sure I could
automate it under mac or win as well, but if it ain't broke...
* Most 'porting' of NQC takes the form of moving the Linux release to
another Unix or Unix-like platform. This is why the source release is
made as a .tar.gz file rather than a .zip or .sit file. Its just one more
reason to tie the linux and source releases together from a process
standpoint.
Overall this means there's a non-trivial 'cost' associated with including
linux and/or source releases in the beta.
As for excluding the non mac/win users from the beta, historically I never
felt this was too big a deal. In the NQC demographic, windows users by
far outnumber all other platforms combined. I've only continued to
release mac as beta since I build it anyway for my own regression tests
(which are done on a mac). In fact, I've even been considering
eliminating the mac beta.
In theory, not releasing source would make bug fixes harder. However,
historically there's not much evidence to suggest a beta source release
would matter. To my recollection no bug has been found during a beta test
period (not counting features which were known up front to be only
partially implemented at beta, etc). There have only been a handful of
actual bugs in NQC itself (in over a year and something like 5 point
releases). In short, the beta test doesn't really shake out many bugs -
its really more of a safety net and allows me to sometimes get new
features (such as the 4x download) tested on different hardware before
releasing.
There also hasn't been a lot of external contribution on bug fixes. This
is by no means a complaint - the documentation in the NQC source is
extermely sparse. If I hadn't written it myself, I'd never try fixing
bugs in it. Bugs have been rare enough, and easy enough to reproduce so
far that there really isn't a need for other people to spend time helping
me debug it.
The area of greatest contribution so far has been getting NQC ported to
other platforms. This has lead to a more portable implementation of
PSerial_unix.cpp, and in general has helped the Makefile evolve a bit. I
don't think beta source releases would impact these porting efforts one
way or another.
As a final note, this latest beta has been running a lot longer than
previous ones, and that perhaps is causing some linux users to be
anxious. The reason its running so long is that around the holidays I was
too busy to do anything with NQC, and lately I've been cramming more
features into it (such as a 'switch' statement). Hopefully, things will
return pretty soon to a much shorter beta cycle and people won't have to
wait so long for linux and source releases.
Dave Baum
--
reply to: dbaum at enteract dot com
|
|
Message has 2 Replies:
 | | Re: NQC beta test release policy
|
| oops.. a slight cock-up on my part means that this message is delivered in installments, read this one as a continuation of the one it is in reply to:-) About the source release being "tied to the linux release": Considering the fact that the linux (...) (25 years ago, 7-Jan-00, to lugnet.robotics.rcx.nqc)
|
Message is in Reply To:
| | NQC beta test release policy
|
| As this is my first post to this newsgroup, I thought I'd start by saying how wonderful I think nqc is etc. I got the RIS 1.5 for xmas and had nqc up and running in no time on my linux system. As well as being an excellent language and bytecode (...) (25 years ago, 6-Jan-00, to lugnet.robotics.rcx.nqc)
|
9 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
