From nordland at csee.ltu.se Tue Feb 20 11:12:07 2007 From: nordland at csee.ltu.se (Johan Nordlander) Date: Tue Feb 20 11:05:51 2007 Subject: [Timber] Compiler milestone Message-ID: Friends, I'm happy to report that the Timber compiler is now in a such a state that I believe it's time to switch focus to testing. All passes are implemented and no crucial functionality is left out. However, very few test runs have been done up to this point, which means that we are still far from having a running compiler. The run- time system also haven't been checked in yet, but it will be in a day or two. Still I'd be happy for some help producing dry-run test examples at this stage, before we risk getting stuck with issues that relate to I/O and Unix interfacing. My suggestion is that we split the testing activities into two phases: 1. Random testing of everyone's favorite code snippets. First goal is to produce C output only, then in a few days also to make them run. This is likely to draw attention to many of the (no doubt) numerous simple and silly bugs that hinder normal operation. While this activity might generate a lot of error reports initially it should hopefully also lead to a relatively well-behaved compiler in a rather short time. I expect to do most of the bug-hunting myself while in this phase. 2. Systematic testing of all branches of every individual pass. This should probably be combined with careful documentation, perhaps also with writing/revising the language definition. To make this work feasible we should split it up in some sensible way, although it's probably wise to wait until phase 1 is over until we make an attempt at organizing phase 2. I also have a list of minor areas of the compiler that still need to be completed, as well as some discussion points regarding the actual language design. But again it's probably best to defer these issues until we at least have a compiler that runs the basic test examples. So if you're interested in the long term goal and prepared for some initial disappointments, pull out the latest darcs version and generate some error reports! Send them directly to me and I'll respond to the list after each update to the sources. Happy lumbering! Johan From leijon at csee.ltu.se Tue Feb 20 11:19:11 2007 From: leijon at csee.ltu.se (Viktor Leijon) Date: Tue Feb 20 11:13:10 2007 Subject: [Timber] Compiler milestone In-Reply-To: References: Message-ID: Hej. Det f?rsta du ska f? ?r min parser-diff: pub98-187:~/tmp/timber/timber leijon$ darcs diff Parser.y 192c192 < : selvar exp { Field $1 $2 } --- > : selvar '=' exp { Field $1 $3 } Jag ?terkommer med testfall som jag tycker borde fungera. /Viktor On 20 feb 2007, at 17.12, Johan Nordlander wrote: > Friends, > > I'm happy to report that the Timber compiler is now in a such a > state that I believe it's time to switch focus to testing. All > passes are implemented and no crucial functionality is left out. > > However, very few test runs have been done up to this point, which > means that we are still far from having a running compiler. The > run-time system also haven't been checked in yet, but it will be in > a day or two. Still I'd be happy for some help producing dry-run > test examples at this stage, before we risk getting stuck with > issues that relate to I/O and Unix interfacing. > > My suggestion is that we split the testing activities into two phases: > > 1. Random testing of everyone's favorite code snippets. First goal > is to produce C output only, then in a few days also to make them > run. This is likely to draw attention to many of the (no doubt) > numerous simple and silly bugs that hinder normal operation. While > this activity might generate a lot of error reports initially it > should hopefully also lead to a relatively well-behaved compiler in > a rather short time. I expect to do most of the bug-hunting myself > while in this phase. > > 2. Systematic testing of all branches of every individual pass. > This should probably be combined with careful documentation, > perhaps also with writing/revising the language definition. To > make this work feasible we should split it up in some sensible way, > although it's probably wise to wait until phase 1 is over until we > make an attempt at organizing phase 2. > > I also have a list of minor areas of the compiler that still need > to be completed, as well as some discussion points regarding the > actual language design. But again it's probably best to defer > these issues until we at least have a compiler that runs the basic > test examples. > > So if you're interested in the long term goal and prepared for some > initial disappointments, pull out the latest darcs version and > generate some error reports! Send them directly to me and I'll > respond to the list after each update to the sources. > > Happy lumbering! > Johan > > _______________________________________________ > Timber mailing list > Timber@haskell.org > http://www.haskell.org/mailman/listinfo/timber From leijon at csee.ltu.se Tue Feb 20 11:24:34 2007 From: leijon at csee.ltu.se (Viktor Leijon) Date: Tue Feb 20 11:18:32 2007 Subject: [Timber] Compiler milestone In-Reply-To: References: Message-ID: Hello all. I missed that reply-to was set, how embarassing. Anyhow, I was simply saying that I had a little diff to the parser, and that I was going to send some test cases later. /Viktor On 20 feb 2007, at 17.19, Viktor Leijon wrote: > Hej. > > Det f?rsta du ska f? ?r min parser-diff: > pub98-187:~/tmp/timber/timber leijon$ darcs diff Parser.y > 192c192 > < : selvar exp { Field $1 $2 } > --- > > : selvar '=' exp { Field $1 > $3 } > > Jag ?terkommer med testfall som jag tycker borde fungera. > > /Viktor > > On 20 feb 2007, at 17.12, Johan Nordlander wrote: > >> Friends, >> >> I'm happy to report that the Timber compiler is now in a such a >> state that I believe it's time to switch focus to testing. All >> passes are implemented and no crucial functionality is left out. >> >> However, very few test runs have been done up to this point, which >> means that we are still far from having a running compiler. The >> run-time system also haven't been checked in yet, but it will be >> in a day or two. Still I'd be happy for some help producing dry- >> run test examples at this stage, before we risk getting stuck with >> issues that relate to I/O and Unix interfacing. >> >> My suggestion is that we split the testing activities into two >> phases: >> >> 1. Random testing of everyone's favorite code snippets. First >> goal is to produce C output only, then in a few days also to make >> them run. This is likely to draw attention to many of the (no >> doubt) numerous simple and silly bugs that hinder normal >> operation. While this activity might generate a lot of error >> reports initially it should hopefully also lead to a relatively >> well-behaved compiler in a rather short time. I expect to do most >> of the bug-hunting myself while in this phase. >> >> 2. Systematic testing of all branches of every individual pass. >> This should probably be combined with careful documentation, >> perhaps also with writing/revising the language definition. To >> make this work feasible we should split it up in some sensible >> way, although it's probably wise to wait until phase 1 is over >> until we make an attempt at organizing phase 2. >> >> I also have a list of minor areas of the compiler that still need >> to be completed, as well as some discussion points regarding the >> actual language design. But again it's probably best to defer >> these issues until we at least have a compiler that runs the basic >> test examples. >> >> So if you're interested in the long term goal and prepared for >> some initial disappointments, pull out the latest darcs version >> and generate some error reports! Send them directly to me and >> I'll respond to the list after each update to the sources. >> >> Happy lumbering! >> Johan >> >> _______________________________________________ >> Timber mailing list >> Timber@haskell.org >> http://www.haskell.org/mailman/listinfo/timber > > _______________________________________________ > Timber mailing list > Timber@haskell.org > http://www.haskell.org/mailman/listinfo/timber > From nordland at csee.ltu.se Wed Feb 21 20:23:55 2007 From: nordland at csee.ltu.se (Johan Nordlander) Date: Wed Feb 21 20:17:34 2007 Subject: [Timber] Compiler milestone In-Reply-To: References: Message-ID: <6D770806-5E21-448F-953E-F652B2CCE7EC@csee.ltu.se> Fixat. -- Johan > Hej. > > Det f?rsta du ska f? ?r min parser-diff: > pub98-187:~/tmp/timber/timber leijon$ darcs diff Parser.y > 192c192 > < : selvar exp { Field $1 $2 } > --- > > : selvar '=' exp { Field $1 > $3 } > > Jag ?terkommer med testfall som jag tycker borde fungera. > > /Viktor > > On 20 feb 2007, at 17.12, Johan Nordlander wrote: > >> Friends, >> >> I'm happy to report that the Timber compiler is now in a such a >> state that I believe it's time to switch focus to testing. All >> passes are implemented and no crucial functionality is left out. >> >> However, very few test runs have been done up to this point, which >> means that we are still far from having a running compiler. The >> run-time system also haven't been checked in yet, but it will be >> in a day or two. Still I'd be happy for some help producing dry- >> run test examples at this stage, before we risk getting stuck with >> issues that relate to I/O and Unix interfacing. >> >> My suggestion is that we split the testing activities into two >> phases: >> >> 1. Random testing of everyone's favorite code snippets. First >> goal is to produce C output only, then in a few days also to make >> them run. This is likely to draw attention to many of the (no >> doubt) numerous simple and silly bugs that hinder normal >> operation. While this activity might generate a lot of error >> reports initially it should hopefully also lead to a relatively >> well-behaved compiler in a rather short time. I expect to do most >> of the bug-hunting myself while in this phase. >> >> 2. Systematic testing of all branches of every individual pass. >> This should probably be combined with careful documentation, >> perhaps also with writing/revising the language definition. To >> make this work feasible we should split it up in some sensible >> way, although it's probably wise to wait until phase 1 is over >> until we make an attempt at organizing phase 2. >> >> I also have a list of minor areas of the compiler that still need >> to be completed, as well as some discussion points regarding the >> actual language design. But again it's probably best to defer >> these issues until we at least have a compiler that runs the basic >> test examples. >> >> So if you're interested in the long term goal and prepared for >> some initial disappointments, pull out the latest darcs version >> and generate some error reports! Send them directly to me and >> I'll respond to the list after each update to the sources. >> >> Happy lumbering! >> Johan >> >> _______________________________________________ >> Timber mailing list >> Timber@haskell.org >> http://www.haskell.org/mailman/listinfo/timber > > _______________________________________________ > Timber mailing list > Timber@haskell.org > http://www.haskell.org/mailman/listinfo/timber > From andy at galois.com Fri Feb 23 17:53:44 2007 From: andy at galois.com (Andy Gill) Date: Fri Feb 23 17:47:25 2007 Subject: [Timber] Compiler milestone In-Reply-To: References: Message-ID: Fantastic! I'll compile with Hpc as tests are added. Any recommend way of adding tests? Andy On Feb 20, 2007, at 8:12 AM, Johan Nordlander wrote: > Friends, > > I'm happy to report that the Timber compiler is now in a such a > state that I believe it's time to switch focus to testing. All > passes are implemented and no crucial functionality is left out. > > However, very few test runs have been done up to this point, which > means that we are still far from having a running compiler. The > run-time system also haven't been checked in yet, but it will be in > a day or two. Still I'd be happy for some help producing dry-run > test examples at this stage, before we risk getting stuck with > issues that relate to I/O and Unix interfacing. > > My suggestion is that we split the testing activities into two phases: > > 1. Random testing of everyone's favorite code snippets. First goal > is to produce C output only, then in a few days also to make them > run. This is likely to draw attention to many of the (no doubt) > numerous simple and silly bugs that hinder normal operation. While > this activity might generate a lot of error reports initially it > should hopefully also lead to a relatively well-behaved compiler in > a rather short time. I expect to do most of the bug-hunting myself > while in this phase. > > 2. Systematic testing of all branches of every individual pass. > This should probably be combined with careful documentation, > perhaps also with writing/revising the language definition. To > make this work feasible we should split it up in some sensible way, > although it's probably wise to wait until phase 1 is over until we > make an attempt at organizing phase 2. > > I also have a list of minor areas of the compiler that still need > to be completed, as well as some discussion points regarding the > actual language design. But again it's probably best to defer > these issues until we at least have a compiler that runs the basic > test examples. > > So if you're interested in the long term goal and prepared for some > initial disappointments, pull out the latest darcs version and > generate some error reports! Send them directly to me and I'll > respond to the list after each update to the sources. > > Happy lumbering! > Johan > > _______________________________________________ > Timber mailing list > Timber@haskell.org > http://www.haskell.org/mailman/listinfo/timber