On 7/14/07, Erick Tryzelaar <[EMAIL PROTECTED]> wrote:
> On 7/14/07, skaller <[EMAIL PROTECTED] > wrote:
> > Still .. to solve the chicken-and-egg problem it's a snowball effect.
> This is how I came about working on fbuild. I wanted to support
this doesn't help address the topic at hand at all, b
hi,
many thanks to you all for your time and thoughts and eyes and ears.
I don't know that I have the time to be able to do much other than be
a pain in the neck, for which I should have been and must apologize.
Given what-all you-all have been saying, I guess I would hazard some
new suggestions
On Sun, 2007-07-15 at 08:25 +0200, Rhythmic Fistman wrote:
> Now dypgen's crashing with a bus error, even though that
> file doesn't look that complicated:
you should probably report that on dypgen and ocaml lists.
Any suggestions what causes this, on OSX 10.4?
Still a stack overflow?
> * H
On 7/14/07, Peter Tanski <[EMAIL PROTECTED]> wrote:
I was thinking about backups for the felix repository. How much space
does SourceForge give you? Maybe you could start nightly build runs (with
buildbot, for example) and when a number of machines pass the regression
tests, use a script-hook
On Sun, 2007-07-15 at 01:38 -0400, Peter Tanski wrote:
> I was thinking about backups for the felix repository. How much space
> does SourceForge give you?
Enough I think: I asked for extra space, so we have a bit
more than the median project.
> Maybe you could start nightly build runs (with
On 7/15/07, skaller <[EMAIL PROTECTED]> wrote:
> On Sun, 2007-07-15 at 11:05 +1000, Rhythmic Fistman wrote:
>
> >
> > Actually it doesn't build out of the box on 32 bit osx and 32 bit windows.
> > Thanks to their smaller stacks and ocaml's penchant for recursion the
> > ocaml crashes compiling flx_
On Sat, 2007-07-14 at 21:28 -0700, Erick Tryzelaar wrote:
> On 7/14/07, skaller <[EMAIL PROTECTED]> wrote:
> This is how I came about working on fbuild. I wanted to support
> out-of-interscript packages, and it was a bit of a pain updating the
> old build system to support this. Of course, then I
On Sat, 2007-07-14 at 21:25 -0700, Erick Tryzelaar wrote:
> On 7/14/07, skaller <[EMAIL PROTECTED]> wrote:
> This is probably going to lead to syntax changes in Felix
> which break things.. better to break them now with few users.
> How long do you think that will take?
Years
On Jul 14, 2007, at 8:14 PM, Erick Tryzelaar wrote:
On 7/14/07, Peter Tanski <[EMAIL PROTECTED]> wrote:
... Darcs gives you the
ability to not tag such files by setting a list of regexes for
'recognised' file names, such as *.c, *.h, *.cpp, *.hpp, *.flx,
*.pak... then you can have files with ot
On 7/14/07, skaller <[EMAIL PROTECTED]> wrote:
Still .. to solve the chicken-and-egg problem it's a snowball effect.
We simply don't have the resources to fix it. We have to hope someone
comes along that is inspired enough by the possibilities to take
on some aspect of the job. If enough people
On 7/14/07, skaller <[EMAIL PROTECTED]> wrote:
On Sat, 2007-07-14 at 17:33 -0700, Erick Tryzelaar wrote:
> (on a side note, skaller, what do you think about releasing another
> felix version once the grammar stuff is stabilized)?
Sure .. but first .. define 'stabilised' .. :)
More likely here
On Sun, 2007-07-15 at 11:05 +1000, Rhythmic Fistman wrote:
>
> Actually it doesn't build out of the box on 32 bit osx and 32 bit windows.
> Thanks to their smaller stacks and ocaml's penchant for recursion the
> ocaml crashes compiling flx_parse.ml.
Try again .. since the grammar rewrite is done
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
> That is - honestly - an impressive list. But... I feel like there is a
> gap between what experienced Felixers perceive as what it can do vs. a
> newbie's experience walking in the door. I don't want to detract from
> what Felix is and does,
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
> You have to see it from the perspective of a comparison shopper who
> sees lots of projects that claim to do wonders but in the end turn out
> not to even build w/out a lot of work.
You should note that *building* Felix is for Felix developer
On Sat, 2007-07-14 at 17:33 -0700, Erick Tryzelaar wrote:
> (on a side note, skaller, what do you think about releasing another
> felix version once the grammar stuff is stabilized)?
Sure .. but first .. define 'stabilised' .. :)
More likely here (as mentioned on 'laziness' I haven't bothered
u
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
>
> sincerely $0.01 and rapidly depreciating, perhaps? :-}
lol ..
> P.S.: Felix *is* cool.
The solve the chicken and egg problem -- bring in two more
developers and lets throw some snowballs.
Since I wrote most of it, it is not very appropr
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
> b) get a sprint happening: physically bring together a bunch of people
> with a bunch of random hardware [modulo (a)] to do full installs from
> scratch, and see how it goes.
You're here aren't you? As a result of LtU posting?
We basically
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
> a) perhaps focus on 2 or 3 core systems and make those work as close
> to flawlessly as you can.
To me, Windows and OSX are 'niche' systems .. however half the
existing developers and many of the new people coming along
are using OSX, and the
On Sat, 2007-07-14 at 16:29 -0700, Raoul Duke wrote:
> > Yes, we do have problems of course .. but there's no way to fix
> > them unless users report them.
>
> You make an excellent point. I wonder how chicken-and-egg it is, tho?
It's very chicken-and-egg. The system bug-rate is largely due to
f
On Sat, 2007-07-14 at 21:12 +0200, Emmanuel Onzon wrote:
> Each action that is added by the user is numbered increasingly,
> and instead of just returning an s-expression it also returns
> its number.
> Then you can write a merge function that selects the AST that is
> coupled with the highest num
> From: "Erick Tryzelaar" <[EMAIL PROTECTED]>
> Thats really pretty much what we end up doing. Ever since sourceforge shut
> down their compile farm we haven't had the ability to build on the more
> esoteric systems (I'm looking at you, solaris!).
Solaris works fine under vmware fusion. I tried to
> From: skaller <[EMAIL PROTECTED]>
> On Sat, 2007-07-14 at 10:46 -0700, Raoul Duke wrote:
>
> > back to the point / quote at the top of this message: having a
> > smoothly running environment in terms of revision control, building,
> > testing and installing - as well as docs, but that's more
> >
On 7/14/07, Raoul Duke <[EMAIL PROTECTED]> wrote:
hi,
> Felix builds 'out of the box' on Unixen, OSX, Cygwin, MingW,
> Win32 and Win64. Do you know ANY other system that can build
> such a complex suite of software -- and test it -- on all
> those platforms? I don't.
That is - honestly - an im
On 7/14/07, Peter Tanski <[EMAIL PROTECTED]> wrote:
Someone also mentioned in the above thread a little about using
emacs, with all those little backup.ext~ files. Darcs gives you the
ability to not tag such files by setting a list of regexes for
'recognised' file names, such as *.c, *.h, *.cpp
hi,
> Felix builds 'out of the box' on Unixen, OSX, Cygwin, MingW,
> Win32 and Win64. Do you know ANY other system that can build
> such a complex suite of software -- and test it -- on all
> those platforms? I don't.
That is - honestly - an impressive list. But... I feel like there is a
gap betw
On Jul 14, 2007, at 3:13 PM, Erick Tryzelaar wrote:
> skaller wrote:
>> On Sat, 2007-07-14 at 14:46 -0400, Peter Tanski wrote:
>>
>>> Why not use Darcs?
>>
>> See previous email from Erick.. Darcs is broken.
>>
>>
>
> In case you missed the conversation, the problem is that darcs can
> fall int
skaller wrote:
> On Sat, 2007-07-14 at 14:46 -0400, Peter Tanski wrote:
>
>> Why not use Darcs?
>>
>
> See previous email from Erick.. Darcs is broken.
>
>
In case you missed the conversation, the problem is that darcs can fall
into a subtle database corruption with very long branche
Quoting skaller <[EMAIL PROTECTED]>:
> On Sat, 2007-07-14 at 20:08 +0200, Emmanuel Onzon wrote:
>> Quoting skaller <[EMAIL PROTECTED]>:
>
>> > Dypgen will then call a merge function which by default
>> > choses the latest definition, our new one.
>>
>> Note that it is not possible to know which AST
On Sun, 2007-07-15 at 03:36 +1000, skaller wrote:
> I have added a new feature to the Felix parser to make this work:
And now this works:
///
open syntax cstuff;
fun f(int x, long *y, z:int):int = {
return x + int(*y) + z;
}
var x = 5L;
println$ f (
On Sat, 2007-07-14 at 14:46 -0400, Peter Tanski wrote:
> Erick Tryzelaar wrote:
> Why not use Darcs?
See previous email from Erick.. Darcs is broken.
--
John Skaller
Felix, successor to C++: http://felix.sf.net
-
This S
Erick Tryzelaar wrote:
> I'd like to do this, some time. The problem is that, pratically
> speaking,
> it's only skaller checking things into the repository. That, and
> branching
> still really sucks in subversion. I'm experimenting with git to see
> how
> effective we could be using that t
On Sat, 2007-07-14 at 20:08 +0200, Emmanuel Onzon wrote:
> Quoting skaller <[EMAIL PROTECTED]>:
> > Dypgen will then call a merge function which by default
> > choses the latest definition, our new one.
>
> Note that it is not possible to know which AST is chosen by the
> default merge function.
On Sat, 2007-07-14 at 10:49 -0700, Raoul Duke wrote:
> > (hm, anybody know why when i try to build my hello.flx all I get is
> > "//Parsing Implementation std.flx" and then it seems to hang for
> > ever?)
>
> it finally moved on, to other issues. i'll infestigate :)
>
> //Parsing Implementation s
On Sat, 2007-07-14 at 10:46 -0700, Raoul Duke wrote:
> back to the point / quote at the top of this message: having a
> smoothly running environment in terms of revision control, building,
> testing and installing - as well as docs, but that's more
> pie-in-the-sky usually - would give a long-term
Quoting skaller <[EMAIL PROTECTED]>:
> Note: if you try something like:
>
> statement := "int" sname ; =># ...
>
> then
>
> int x;
>
> will also be parsed as
>
> call int x;
>
> [that is, as a call of procedure 'int' on argument 'x']
>
> Dypgen will then call a merge function whi
On Sat, 2007-07-14 at 10:46 -0700, Raoul Duke wrote:
> (hm, anybody know why when i try to build my hello.flx all I get is
> "//Parsing Implementation std.flx" and then it seems to hang for
> ever?)
Forever? Or just slow? It takes a while to parse the standard
library, but not forever. Note it is
> (hm, anybody know why when i try to build my hello.flx all I get is
> "//Parsing Implementation std.flx" and then it seems to hang for
> ever?)
it finally moved on, to other issues. i'll infestigate :)
//Parsing Implementation std.flx
//Parsing Implementation hello.flx
In file included from hel
> Moving the repository, changing the build system, etc etc etc
> are all peripheral destabilising events. We actually need docs,
> performance measurements, and perhaps an easier way for people
> to add regression tests.
hi,
i'll throw in my $0.02 unbidden? :-} [i'm a software developer by
trade
I have added a new feature to the Felix parser to make this work:
#import
syntax csyn {
statement := typedef "int" sname; =># """`(ast_nop ,_sr "int")""";
statement := typedef "long" sname; =># """`(ast_nop ,_sr "long")""";
}
open syntax csyn;
typed
On Fri, 2007-07-13 at 22:50 -0700, Erick Tryzelaar wrote:
> On 7/13/07, skaller <[EMAIL PROTECTED]> wrote:
> That, still, isn't really a backup policy though.
[]
The thing is whether it is worth doing any of these changes.
The priority for the project is to get more people involved.
I can actual
40 matches
Mail list logo
