From patc at cse.unsw.edu.au  Mon Sep  4 22:50:36 2006
From: patc at cse.unsw.edu.au (Patrick Caldon)
Date: Sun Sep 10 12:03:23 2006
Subject: [Hat] Feature request/wish list - library support
Message-ID: <20060905025036.GA24558@cse.unsw.EDU.AU>


Hi,

I don't know how you update your web pages and so on, but I'd like to
make a feature request (wish!) for support of all the standard haskell
libraries.  The couple of times I've tried using Hat this has been an
issue. ST related libraries are the ones I really want.

Also the first time I tried using Hat (this was about a week after I
started using Haskell) I got desparately confused by its lack of
library support.  If I recall, I had a little bit of toy code which
used Data.Queue, and tried re-installing Hat several times with
different options, each time getting an erro message like "Data.Queue:
Does not exist".  Could you change the error message to read?

Data.Queue: Does not exist; check http://<something> for a list of
supported libraries.

Thanks all the same for all your work!

Regards,
Patrick.
From cb224 at cs.york.ac.uk  Sat Sep  9 07:30:42 2006
From: cb224 at cs.york.ac.uk (cb224@cs.york.ac.uk)
Date: Sun Sep 10 12:03:25 2006
Subject: [Hat] building hat with ghc 6.4.2 under debian testing/unstable
Message-ID: <54693.217.37.73.202.1157801442.squirrel@www.cs.york.ac.uk>

hello,

cpp from gcc 4.1.2 appears to output more whitespace than expected when
generating ghcsym.out. this results in the following error during build:

~/src/other/hat-2.04.orig$ make
src/other/hat-2.04.orig/lib/ix86-Linux/config:11: *** missing separator. 
Stop.

the attached patch appears to fix this.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hat-conf-ghc6.4.2.patch
Type: text/x-patch
Size: 1029 bytes
Desc: not available
Url : http://www.haskell.org//pipermail/hat/attachments/20060909/c702498c/hat-conf-ghc6.4.2.bin
From Malcolm.Wallace at cs.york.ac.uk  Mon Sep 11 06:49:09 2006
From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace)
Date: Mon Sep 11 06:36:36 2006
Subject: [Hat] Feature request/wish list - library support
In-Reply-To: <20060905025036.GA24558@cse.unsw.EDU.AU>
References: <20060905025036.GA24558@cse.unsw.EDU.AU>
Message-ID: <20060911114909.15570935.Malcolm.Wallace@cs.york.ac.uk>

patc@cse.unsw.edu.au (Patrick Caldon) wrote:

> I don't know how you update your web pages and so on, but I'd like to
> make a feature request (wish!) for support of all the standard haskell
> libraries.  The couple of times I've tried using Hat this has been an
> issue. ST related libraries are the ones I really want.

Well, Hat does indeed support all the officially "standard" Haskell
libraries, namely those defined by the Haskell'98 Report.  It
additionally supports many of the non-standard libraries contained in
the hierarchical "base" package, but as you discovered, not all of them.

There are good reasons why some (e.g. ST) are not supported - because
they are not portable between compilers.

However, there are plenty of newer non-standard libraries (e.g.
Data.Queue) which are entirely portable, and we have no good excuse for
failing to support them in Hat, except a lack of developer time to chase
them up.

In fact the real failure here is the lack of an easy mechanism for
anyone to add a newer library into Hat's default set.  Automation would
make this so much less of a problem.  But sadly, I do not forsee this
happening in the near future.

> Could you change the error message to read?
> 
> Data.Queue: Does not exist;
>   check http://<something> for a list of supported libraries.

A great idea.

Regards,
    Malcolm
From Malcolm.Wallace at cs.york.ac.uk  Mon Sep 11 06:51:09 2006
From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace)
Date: Mon Sep 11 06:45:25 2006
Subject: [Hat] building hat with ghc 6.4.2 under debian testing/unstable
In-Reply-To: <54693.217.37.73.202.1157801442.squirrel@www.cs.york.ac.uk>
References: <54693.217.37.73.202.1157801442.squirrel@www.cs.york.ac.uk>
Message-ID: <20060911115109.50127dbf.Malcolm.Wallace@cs.york.ac.uk>

cb224@cs.york.ac.uk wrote:

> cpp from gcc 4.1.2 appears to output more whitespace than expected
> when generating ghcsym.out. this results in the following error

Thanks for the bug report.  A patch for this already exists on the Hat
download page.  (Patch name: "config")

Regards,
    Malcolm
From John.DeTreville at microsoft.com  Fri Sep 15 19:55:02 2006
From: John.DeTreville at microsoft.com (John DeTreville)
Date: Fri Sep 15 20:30:44 2006
Subject: [Hat] Documentation on trace file format?
Message-ID: <CE46CB2C027EF247B3924761571C41A40452F828@NA-EXMSG-C108.redmond.corp.microsoft.com>

Hello,

I'm in the middle of a small research project that analyzes execution traces of large Haskell programs. I'm currently using Hat as an infrastructure for generating traces (same hat-trans, same libraries, same trace files, custom back-end).

I'd first like to thanks the Hat authors and maintainers for their fine work. I've tried or considered a few other ways to generate such traces, and Hat provides a great combination of broad language and library coverage with accurate traces and the ability to trace long executions. Thank you!

My biggest problem (at the moment) is that the trace file format is under-documented. I've written my new back-end by reverse-engineering existing back-ends, but that's too error-prone for comfort. Is there any unpublished documentation on the trace file format and its correct interpretation that might help me ensure I'm not missing anything crucial?

Cheers,
John
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.haskell.org/pipermail/hat/attachments/20060915/ffb2fc44/attachment-0001.htm
From malcolm at cs.york.ac.uk  Fri Sep 15 21:01:25 2006
From: malcolm at cs.york.ac.uk (malcolm@cs.york.ac.uk)
Date: Sat Sep 16 17:28:42 2006
Subject: [Hat] Documentation on trace file format?
In-Reply-To: <CE46CB2C027EF247B3924761571C41A40452F828@NA-EXMSG-C108.redmond.corp.m
	icrosoft.com>
References: <CE46CB2C027EF247B3924761571C41A40452F828@NA-EXMSG-C108.redmond.corp.microsoft.com>
Message-ID: <54488.131.252.226.113.1158368485.squirrel@www.cs.york.ac.uk>

> My biggest problem (at the moment) is that the trace file format is
> under-documented. I've written my new back-end by reverse-engineering
> existing back-ends, but that's too error-prone for comfort. Is there any
> unpublished documentation on the trace file format and its correct
> interpretation that might help me ensure I'm not missing anything crucial?

Indeed, there is a lot of unpublished documentation about Hat.  We had a
series of internal memos covering lots of design issues including the file
format.  Up until now, these have been private, but I can't think of a
good reason for them to remain that way, especially if other people are
developing on top of Hat and need to understand such issues.

So have a look at
    http://www.cs.york.ac.uk/fp/hat/Memos

Bearing in mind that many of the memos may now be superseded or out of
date, I hope that nevertheless you manage to find something useful here. 
(There are two separate memos about the file format, one by me, another by
Olaf.)

Regards,
    Malcolm

From ndmitchell at gmail.com  Thu Sep 28 20:47:23 2006
From: ndmitchell at gmail.com (Neil Mitchell)
Date: Thu Sep 28 19:53:28 2006
Subject: [Hat] Parsing issue with 1e12
Message-ID: <404396ef0609281747p13af203fm96df2d3aa5a14170@mail.gmail.com>

Hi Hat and Yhc people (and I guess nhc people as well),

Both GHC and Hugs parse 1e12 as a number, but Hat and Yhc don't.

For example, test = print 1e12 gives:
yhc: -- during when renaming
Error: Identifier e12 used at 7:15-7:17 is not defined.

I have also been told that Hat leads to an error.

So:

1) Is this an error, what does the Haskell spec say?

2) Should the parser be fixed in all the right places? (does this
impact Colin's hpc as well?)

Thanks

Neil
From Malcolm.Wallace at cs.york.ac.uk  Fri Sep 29 04:24:37 2006
From: Malcolm.Wallace at cs.york.ac.uk (Malcolm Wallace)
Date: Fri Sep 29 03:30:40 2006
Subject: [Hat] Re: Parsing issue with 1e12
In-Reply-To: <404396ef0609281747p13af203fm96df2d3aa5a14170@mail.gmail.com>
References: <404396ef0609281747p13af203fm96df2d3aa5a14170@mail.gmail.com>
Message-ID: <20060929092437.522f1a97.Malcolm.Wallace@cs.york.ac.uk>

"Neil Mitchell" <ndmitchell@gmail.com> writes:

> Both GHC and Hugs parse 1e12 as a number, but Hat and Yhc don't.

I suspect this was an unnoticed change in the Haskell Report spec.
Originally I think the exponent form was allowed only if there was a
decimal point before the 'e'.  I see this is no longer the case.
Yes, this is likely to affect the hpc parser as well.

Regards,
    Malcolm