Re: [sqlite] Fossil bugs?
On Thu, May 19, 2011 at 1:40 PM, Nico Williams wrote: > On Thu, May 19, 2011 at 12:37 PM, Richard Hipp wrote: > > On Thu, May 19, 2011 at 1:23 PM, Nico Williams >wrote: > > Everything is protected by multiple cryptographic hashes, both SHA1 and > > MD5. On-the-wire corruption is not a realistic possibility. > > Excellent. > > > Is that normal? The missing whitespace between '{*}' and '$result' > >> causes make test failures. > > > > There is not suppose to be a space between the {*} and the $. The > {*}$var > > is a relatively new construct in TCL that means to expand the variable as > > one or more space-separated arguments. It is a safer alternative to the > old > > "eval" command. > > So I need a new version of Tcl? I'm using 8.4. What version do I need? > I think 8.5 is sufficient. But if you are upgrading, you might as well go with the very latest 8.6 beta. > > Thanks, > > Nico > -- > ___ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] Fossil bugs?
On Thu, May 19, 2011 at 12:37 PM, Richard Hipp wrote: > On Thu, May 19, 2011 at 1:23 PM, Nico Williams wrote: > Everything is protected by multiple cryptographic hashes, both SHA1 and > MD5. On-the-wire corruption is not a realistic possibility. Excellent. > Is that normal? The missing whitespace between '{*}' and '$result' >> causes make test failures. > > There is not suppose to be a space between the {*} and the $. The {*}$var > is a relatively new construct in TCL that means to expand the variable as > one or more space-separated arguments. It is a safer alternative to the old > "eval" command. So I need a new version of Tcl? I'm using 8.4. What version do I need? Thanks, Nico -- ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] Fossil bugs?
On Thu, May 19, 2011 at 1:23 PM, Nico Williams wrote: > On Thu, May 19, 2011 at 12:01 PM, Richard Hipp wrote: > > Anythings possible. But we've been using Fossil heavily, daily, for 4 > years > > now without any hints of these kinds of problems. So fundamental bugs > like > > what you propose seem improbable. I'm thinking something else is going > on. > > Is TLS used? If not, is data sent over HTTP integrity protected in > any way (e.g., with hashes)? If not, then the tcl script errors are > probably due to on-the-wire corruption. > Everything is protected by multiple cryptographic hashes, both SHA1 and MD5. On-the-wire corruption is not a realistic possibility. Is that normal? The missing whitespace between '{*}' and '$result' > causes make test failures. > There is not suppose to be a space between the {*} and the $. The {*}$var is a relatively new construct in TCL that means to expand the variable as one or more space-separated arguments. It is a safer alternative to the old "eval" command. -- D. Richard Hipp d...@sqlite.org ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] Fossil bugs?
On Thu, May 19, 2011 at 12:01 PM, Richard Hipp wrote: > Anythings possible. But we've been using Fossil heavily, daily, for 4 years > now without any hints of these kinds of problems. So fundamental bugs like > what you propose seem improbable. I'm thinking something else is going on. Is TLS used? If not, is data sent over HTTP integrity protected in any way (e.g., with hashes)? If not, then the tcl script errors are probably due to on-the-wire corruption. However, I just checked the tarball for 3.7.6.2 and it also has those missing whitespaces in Tcl files, like this: proc do_select_test {name sql result} { uplevel [list doPassiveTest 0 $name $sql [list 0 [list {*}$result]]] } which I think rules out corruption on the network (I'd almost certainly not get the same corruption twice, once in a tarball and once in Fossil clone). Is that normal? The missing whitespace between '{*}' and '$result' causes make test failures. > Like maybe your text editor is automatically converting space into tabs? I hadn't edited any of these files, so, no. > What platform did you say you were running on? Our 4-years of experience is > mostly confined to Linux/Mac/Windows. Are you doing this on Solaris or > something? Maybe you have an NFS-mounted home directory? Something else > out of the ordinary? Ubuntu 10.1, in a local directory. But inside a Virtual Box VM -- but I've not noticed corruption problems before. I'm also using wifi (see comments above about integrity protection. I did a new "fossil open" of my clone of sqlite3docs and this time I got a copy of Makefile checked out. In my previous workspace I removed the Makefile, did a fossil commit --latest --force, and now everything looks good. I don't know what caused the problem I had with the Makefile -- let's blame it on user error until it happens again. Nico -- ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] Fossil bugs?
On Thu, May 19, 2011 at 11:58 AM, Nico Williams wrote: > I installed Fossil using aptitude, but I'm thinking I should have > built and installed it from source. I'm seeing a few issues with the > version I'm using ([15cb835736] 2010-06-17 18:39:10 UTC): > > - The Makefile from the SQLite3 docs repository doesn't get checked > out -- I thought there wasn't any, but "fossil diff" shows changes to > it, and "fossil undo" and "fossil revert" do not restore the original. > This strikes me as almost certainly a bug. > > - In the SQLite3 repository I had to make a bunch of changes like this > one: > > - uplevel do_test ${prefix}.$tn [list $tclquery] [list [list {*}$res]] > + uplevel do_test ${prefix}.$tn [list $tclquery] [list [list {*} > $res]] > > I worry that the missing whitespace may be a result of a bug in Fossil. > Anythings possible. But we've been using Fossil heavily, daily, for 4 years now without any hints of these kinds of problems. So fundamental bugs like what you propose seem improbable. I'm thinking something else is going on. Like maybe your text editor is automatically converting space into tabs? What platform did you say you were running on? Our 4-years of experience is mostly confined to Linux/Mac/Windows. Are you doing this on Solaris or something? Maybe you have an NFS-mounted home directory? Something else out of the ordinary? > > I will try a newer version of Fossil. > > Nico > -- > ___ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users