License: Public Domain
YMMV
As of 6/25/2015, I could not get Clasp to compile, even once. See #clasp IRC log with entries tagged "clasp-build:". Ignore ones with arobin. I started on 6/20/2015. For me, 6 days is enough for this incarnation of Clasp.
Currently, compiling Clasp from source is not for the faint of heart. Some have done it, others like me have failed. You have to go to an IRC chat room #clasp (admittedly, a nice, pleasant one ;) to do this for all practical purposes. Don't bug Chris unless you are serious because he is busy. I learned today the source code might change underneath you between builds--the make system seems to pull working repositories behind your back each time you run "make". I definitely am not being critical--I have bootstrapped some scary systems, like a prototype VM on a custom PowerPC board--and it isn't easy.
I thought I could do it because: o I have 25 years experience with Lisp and C++ and every language imaginable, in many application areas o I have distributed and contributed to a number of free projects, including building some iffy prototypes o I have worked on an Ada -> custom VM (human) safety critical compiler written from scratch--we weren't allowed to HAVE a make system for reproducibility o I have a strong need for the technology Chris is developing o I am very independent--I don't really even have a hand to hold, being a bird and all o I have a bird-brain and a degree in CS
Where I went wrong: o Believed something I read on the internet o I don't possess knowledge of some of the more advanced concepts in Lisp, such as atoms, cons cells, and "groveling". o I fell for the line of "try this test version". It is (semi) stable for me or nothing from now on. o I have no knowledge of bjam. To be fair (to me) I have built boost (why does a template programming library need to be built?) from source. o I have always been more of a humorist than a software developer--I was born in Missouri and have read all of Mark Twain's works, more than once. Most people don't think I'm funny. o I don't work well with others o I have a bird-brain and a degree in CS
I would resign, but I don't work here, contrary to what my current boss thinks. I guess I can lurk here on the mailing list--unless I'm banned.
Peace, and good luck. :-|
--Andy (o>
Hi Andy,
Just another datapoint, I've had a lot of trouble building the 'newbuild' branch of clasp, *until recently*.
With the past few commits, the build process has run into no hiccups (not even needing a restart in the middle, as was needed in the past), and *has finished with no errors*.
I would try one more time on the most recent (as of writing this) commit of the newbuild branch: 52d51fef56 https://github.com/drmeister/clasp/commit/52d51fef563447f3693fed1c315b3d31e047933e .
Here's some information about my setup/process:
- Arch linux
- use preview branch of externals-clasp: 810eaaf8af https://github.com/drmeister/externals-clasp/commit/810eaaf8af7f845addfa44da3edbc7a5f82207d0 - clone externals-clasp, checkout preview, copy local.config.linux to local.config & edit configuration, make - more explicitly -- for anyone else (w/o the 25 years of experience) following along:
git clone https://github.com/drmeister/externals-clasp cd externals-clasp git checkout 810eaa cp local.config.linux local.config # edit local.config (in my case, set PJOBS higher, set GCC_TOOLCHAIN = /usr) make
- use newbuild branch of clasp: 52d51fef56 https://github.com/drmeister/clasp/commit/52d51fef563447f3693fed1c315b3d31e047933e - clone clasp, checkout newbuild, copy local.config.linux to local.config & edit configuration, make - more explicitly
git clone https://github.com/drmeister/clasp cd clasp git checkout 52d51f cp local.config.linux local.config # edit local.config (in my case, set EXTERNALS_SOURCE_DIR, set PJOBS higher) make
--- Cheers, Cayman
On Thu, Jun 25, 2015 at 12:27 PM, Andrew Robin andrew.t.robin@gmail.com wrote:
License: Public Domain
YMMV
As of 6/25/2015, I could not get Clasp to compile, even once. See #clasp IRC log with entries tagged "clasp-build:". Ignore ones with arobin. I started on 6/20/2015. For me, 6 days is enough for this incarnation of Clasp.
Currently, compiling Clasp from source is not for the faint of heart. Some have done it, others like me have failed. You have to go to an IRC chat room #clasp (admittedly, a nice, pleasant one ;) to do this for all practical purposes. Don't bug Chris unless you are serious because he is busy. I learned today the source code might change underneath you between builds--the make system seems to pull working repositories behind your back each time you run "make". I definitely am not being critical--I have bootstrapped some scary systems, like a prototype VM on a custom PowerPC board--and it isn't easy.
I thought I could do it because: o I have 25 years experience with Lisp and C++ and every language imaginable, in many application areas o I have distributed and contributed to a number of free projects, including building some iffy prototypes o I have worked on an Ada -> custom VM (human) safety critical compiler written from scratch--we weren't allowed to HAVE a make system for reproducibility o I have a strong need for the technology Chris is developing o I am very independent--I don't really even have a hand to hold, being a bird and all o I have a bird-brain and a degree in CS
Where I went wrong: o Believed something I read on the internet o I don't possess knowledge of some of the more advanced concepts in Lisp, such as atoms, cons cells, and "groveling". o I fell for the line of "try this test version". It is (semi) stable for me or nothing from now on. o I have no knowledge of bjam. To be fair (to me) I have built boost (why does a template programming library need to be built?) from source. o I have always been more of a humorist than a software developer--I was born in Missouri and have read all of Mark Twain's works, more than once. Most people don't think I'm funny. o I don't work well with others o I have a bird-brain and a degree in CS
I would resign, but I don't work here, contrary to what my current boss thinks. I guess I can lurk here on the mailing list--unless I'm banned.
Peace, and good luck. :-|
--Andy (o>
Cayman,
Thanks for that data point and the very explicit directions.
Clasp is at an uncomfortable time in it’s development because it’s been a couple of months since the last release and I’ve made major changes to how it works. I’ve created a lot of loose ends and broken some things (MPS support) and I’m working as fast as I can to fix things and get things to be more stable.
(1) I’ve incorporated tagged pointers and immediate fixnums (2) I’ve added an additional compiler called Cleavir that will enable Clasp to generate fast code that I think can compete with any other Common Lisp compiler. (3) I’ve simplified the build process.
Note the points above - they all start with “I’ve” - yep, it’s just one person doing all of this while managing a lab of 10 people and writing grant proposals.
I would love help but in the immediate future it’s easier for me to just finish up the loose ends that I’ve created, and push out a new release. It would take longer to explain what I’m doing than to just do it.
Best,
.Chris.
On Jun 25, 2015, at 5:47 PM, Cayman Nava caymannava@gmail.com wrote:
Hi Andy,
Just another datapoint, I've had a lot of trouble building the 'newbuild' branch of clasp, until recently.
With the past few commits, the build process has run into no hiccups (not even needing a restart in the middle, as was needed in the past), and has finished with no errors.
I would try one more time on the most recent (as of writing this) commit of the newbuild branch: 52d51fef56.
Here's some information about my setup/process:
Arch linux
use preview branch of externals-clasp: 810eaaf8af
- clone externals-clasp, checkout preview, copy local.config.linux to local.config & edit configuration, make
- more explicitly -- for anyone else (w/o the 25 years of experience) following along:
git clone https://github.com/drmeister/externals-clasp cd externals-clasp git checkout 810eaa cp local.config.linux local.config # edit local.config (in my case, set PJOBS higher, set GCC_TOOLCHAIN = /usr) make
- use newbuild branch of clasp: 52d51fef56
- clone clasp, checkout newbuild, copy local.config.linux to local.config & edit configuration, make
- more explicitly
git clone https://github.com/drmeister/clasp cd clasp git checkout 52d51f cp local.config.linux local.config # edit local.config (in my case, set EXTERNALS_SOURCE_DIR, set PJOBS higher) make
Cheers, Cayman
On Thu, Jun 25, 2015 at 12:27 PM, Andrew Robin andrew.t.robin@gmail.com wrote: License: Public Domain
YMMV
As of 6/25/2015, I could not get Clasp to compile, even once. See #clasp IRC log with entries tagged "clasp-build:". Ignore ones with arobin. I started on 6/20/2015. For me, 6 days is enough for this incarnation of Clasp.
Currently, compiling Clasp from source is not for the faint of heart. Some have done it, others like me have failed. You have to go to an IRC chat room #clasp (admittedly, a nice, pleasant one ;) to do this for all practical purposes. Don't bug Chris unless you are serious because he is busy. I learned today the source code might change underneath you between builds--the make system seems to pull working repositories behind your back each time you run "make". I definitely am not being critical--I have bootstrapped some scary systems, like a prototype VM on a custom PowerPC board--and it isn't easy.
I thought I could do it because: o I have 25 years experience with Lisp and C++ and every language imaginable, in many application areas o I have distributed and contributed to a number of free projects, including building some iffy prototypes o I have worked on an Ada -> custom VM (human) safety critical compiler written from scratch--we weren't allowed to HAVE a make system for reproducibility o I have a strong need for the technology Chris is developing o I am very independent--I don't really even have a hand to hold, being a bird and all o I have a bird-brain and a degree in CS
Where I went wrong: o Believed something I read on the internet o I don't possess knowledge of some of the more advanced concepts in Lisp, such as atoms, cons cells, and "groveling". o I fell for the line of "try this test version". It is (semi) stable for me or nothing from now on. o I have no knowledge of bjam. To be fair (to me) I have built boost (why does a template programming library need to be built?) from source. o I have always been more of a humorist than a software developer--I was born in Missouri and have read all of Mark Twain's works, more than once. Most people don't think I'm funny. o I don't work well with others o I have a bird-brain and a degree in CS
I would resign, but I don't work here, contrary to what my current boss thinks. I guess I can lurk here on the mailing list--unless I'm banned.
Peace, and good luck. :-| --Andy (o>
Andrew,
This sort of thing is bound to happen because Clasp is under active development. You probably had problems building Clasp because the build system was (and may still be) broken. I’m working on it.
I’m going to put out a new release in the next couple of weeks that will incorporate a whole new compiler and simplify the build process by eliminating externals-clasp.
Cheers,
.Chris.
On Jun 25, 2015, at 3:27 PM, Andrew Robin andrew.t.robin@gmail.com wrote:
License: Public Domain
YMMV
As of 6/25/2015, I could not get Clasp to compile, even once. See #clasp IRC log with entries tagged "clasp-build:". Ignore ones with arobin. I started on 6/20/2015. For me, 6 days is enough for this incarnation of Clasp.
Currently, compiling Clasp from source is not for the faint of heart. Some have done it, others like me have failed. You have to go to an IRC chat room #clasp (admittedly, a nice, pleasant one ;) to do this for all practical purposes. Don't bug Chris unless you are serious because he is busy. I learned today the source code might change underneath you between builds--the make system seems to pull working repositories behind your back each time you run "make". I definitely am not being critical--I have bootstrapped some scary systems, like a prototype VM on a custom PowerPC board--and it isn't easy.
I thought I could do it because: o I have 25 years experience with Lisp and C++ and every language imaginable, in many application areas o I have distributed and contributed to a number of free projects, including building some iffy prototypes o I have worked on an Ada -> custom VM (human) safety critical compiler written from scratch--we weren't allowed to HAVE a make system for reproducibility o I have a strong need for the technology Chris is developing o I am very independent--I don't really even have a hand to hold, being a bird and all o I have a bird-brain and a degree in CS
Where I went wrong: o Believed something I read on the internet o I don't possess knowledge of some of the more advanced concepts in Lisp, such as atoms, cons cells, and "groveling". o I fell for the line of "try this test version". It is (semi) stable for me or nothing from now on. o I have no knowledge of bjam. To be fair (to me) I have built boost (why does a template programming library need to be built?) from source. o I have always been more of a humorist than a software developer--I was born in Missouri and have read all of Mark Twain's works, more than once. Most people don't think I'm funny. o I don't work well with others o I have a bird-brain and a degree in CS
I would resign, but I don't work here, contrary to what my current boss thinks. I guess I can lurk here on the mailing list--unless I'm banned.
Peace, and good luck. :-| --Andy (o>
-
Andrew Robin -
Cayman Nava -
Christian Schafmeister