Thread: Demo ideas?
Folks, Next week I need to "demo" PostgreSQL to a tech analyst. Presumably we'll be using some sort of remote desktop into one of my machines. Does anyone have suggestions on some quick things I can do to show off some of PostgreSQL's unique features? --Josh
On Thu, Feb 02, 2006 at 09:58:38AM -0800, Josh Berkus wrote: > Folks, > > Next week I need to "demo" PostgreSQL to a tech analyst. Presumably > we'll be using some sort of remote desktop into one of my machines. > > Does anyone have suggestions on some quick things I can do to show > off some of PostgreSQL's unique features? PL/R and PostGIS would be good for showing off extensibility. I'd mention DBI-Link, but I'm not a disinterested party ;) Cheers, D -- David Fetter david@fetter.org http://fetter.org/ phone: +1 415 235 3778 Remember to vote!
Would an MPP demo be interesting? A 1TB dataset could be available...
- Luke
On 2/2/06 9:58 AM, "Josh Berkus" <josh@agliodbs.com> wrote:
- Luke
On 2/2/06 9:58 AM, "Josh Berkus" <josh@agliodbs.com> wrote:
Folks,
Next week I need to "demo" PostgreSQL to a tech analyst. Presumably
we'll be using some sort of remote desktop into one of my machines.
Does anyone have suggestions on some quick things I can do to show off
some of PostgreSQL's unique features?
--Josh
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
On Thu, Feb 02, 2006 at 10:05:49AM -0800, David Fetter wrote: > On Thu, Feb 02, 2006 at 09:58:38AM -0800, Josh Berkus wrote: > > Does anyone have suggestions on some quick things I can do to show > > off some of PostgreSQL's unique features? > > PL/R and PostGIS would be good for showing off extensibility. I'd > mention DBI-Link, but I'm not a disinterested party ;) I was going to mention PostGIS as well. Showing that you can write server-side functions in many languages (Perl, Python, Tcl, Ruby, PHP, R, sh, etc.) should score geek points, especially if you demonstrate what can be done with set-returning functions. -- Michael Fuhr
On Thu, 2006-02-02 at 13:36, Michael Fuhr wrote:
> On Thu, Feb 02, 2006 at 10:05:49AM -0800, David Fetter wrote:
> > On Thu, Feb 02, 2006 at 09:58:38AM -0800, Josh Berkus wrote:
> > > Does anyone have suggestions on some quick things I can do to show
> > > off some of PostgreSQL's unique features?
> >
> > PL/R and PostGIS would be good for showing off extensibility. I'd
> > mention DBI-Link, but I'm not a disinterested party ;)
>
> I was going to mention PostGIS as well. Showing that you can write
> server-side functions in many languages (Perl, Python, Tcl, Ruby,
> PHP, R, sh, etc.) should score geek points, especially if you
> demonstrate what can be done with set-returning functions.
>
Another option would be to show them pgAdmin III working around a slony
cluster. While *I* think writing function in multiple languages is cool,
it's not necessarily GUI friendly. Judge your audience accordingly.
Robert Treat
--
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
On Thu, Feb 02, 2006 at 11:36:43AM -0700, Michael Fuhr wrote: > On Thu, Feb 02, 2006 at 10:05:49AM -0800, David Fetter wrote: > > On Thu, Feb 02, 2006 at 09:58:38AM -0800, Josh Berkus wrote: > > > Does anyone have suggestions on some quick things I can do to > > > show off some of PostgreSQL's unique features? > > > > PL/R and PostGIS would be good for showing off extensibility. I'd > > mention DBI-Link, but I'm not a disinterested party ;) > > I was going to mention PostGIS as well. Showing that you can write > server-side functions in many languages (Perl, Python, Tcl, Ruby, > PHP, R, sh, etc.) should score geek points, especially if you > demonstrate what can be done with set-returning functions. I mentioned PL/R and PostGIS because they both have easy-to-see visual demonstrations. PostGIS is obvious, but maybe PL/R wasn't... http://www.varlena.com/varlena/GeneralBits/Tidbits/bernier/art_66/graphingWithR.html Cheers, D -- David Fetter david@fetter.org http://fetter.org/ phone: +1 415 235 3778 Remember to vote!
Michael Fuhr wrote: > On Thu, Feb 02, 2006 at 10:05:49AM -0800, David Fetter wrote: >> On Thu, Feb 02, 2006 at 09:58:38AM -0800, Josh Berkus wrote: >>> Does anyone have suggestions on some quick things I can do to show >>> off some of PostgreSQL's unique features? >> PL/R and PostGIS would be good for showing off extensibility. I'd >> mention DBI-Link, but I'm not a disinterested party ;) > > I was going to mention PostGIS as well. Showing that you can write > server-side functions in many languages (Perl, Python, Tcl, Ruby, > PHP, R, sh, etc.) should score geek points, especially if you > demonstrate what can be done with set-returning functions. > Tech analysts seems to have an increasing interest in Java these days. Enterprise features, J2EE, Eclipse, jada jada. Show Eclipse with Clay (http://www.azzurri.jp/en/software/clay/index.jsp) for instance. Easy to set up and gives a good graphical demo. PostgreSQL has an excellent JDBC driver suite (will be used in the demo). And don't forget to mention PL/Java ;-) Regards, Thomas Hallgren
Thomas, > And don't forget to mention > PL/Java ;-) You'll be happy to know that the analyst had a specific question about java procedural support. BTW, what happened to the discussion of moving PL/java into core for 8.2? -- --Josh Josh Berkus Aglio Database Solutions San Francisco
DDL rollback Josh Berkus wrote: > Folks, > > Next week I need to "demo" PostgreSQL to a tech analyst. Presumably > we'll be using some sort of remote desktop into one of my machines. > > Does anyone have suggestions on some quick things I can do to show off > some of PostgreSQL's unique features? > > --Josh > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match
Josh Berkus wrote: > Thomas, > >> And don't forget to mention >> PL/Java ;-) > > You'll be happy to know that the analyst had a specific question about java > procedural support. > > BTW, what happened to the discussion of moving PL/java into core for 8.2? > It got stalled here: http://archives.postgresql.org/pgsql-hackers/2005-08/msg00482.php Regards, Thomas Hallgren
On Feb 2, 2006, at 12:58 PM, Josh Berkus wrote: > Folks, > > Next week I need to "demo" PostgreSQL to a tech analyst. > Presumably we'll be using some sort of remote desktop into one of > my machines. > Does anyone have suggestions on some quick things I can do to show > off some of PostgreSQL's unique features? > You could do a crude datafeed kind of importy thing. Say you're going to import news articles from src x. You're going to get a timestamp, summary and article. Make a type, then a simple perl function to remotely fetch it and return it. I know I've impressed some folks with that. And mr. Fetter's useful dbi-link is nifty too then again, I guess that is more of a geek demo than a PR demo.. geeks get excited by very different things :) That R thing someone posted would be nifty. In fact, I'd never considered doing that before and I may spend the afternoon changing some graphing stuff around. -- Jeff Trout <jeff@jefftrout.com> http://www.jefftrout.com/ http://www.stuarthamm.net/
On Thu, Feb 02, 2006 at 12:29:39PM -0800, Josh Berkus wrote: > Thomas, > > > And don't forget to mention > > PL/Java ;-) > > You'll be happy to know that the analyst had a specific question about java > procedural support. > > BTW, what happened to the discussion of moving PL/java into core for 8.2? > > -- > --Josh > Josh, We tried to include PL/java in our standard build, but the version of the zlib library in the JVM is ancient and the incompatibility would cause the DB to crash. Until we can get a JVM built against a current library from Sun I think it would be a bad idea to have PL/java in the core. Ken
Ken, There are alternatives. The most obvious is perhaps to use gcj and avoid the jvm dependency altogether. Other alternatives are listed under the heading "zlib conflict" here: http://gborg.postgresql.org/project/pljava/genpage.php?readme Question: When is zlib used in the backend today? Kind regards, Thomas Hallgren Kenneth Marshall wrote: > On Thu, Feb 02, 2006 at 12:29:39PM -0800, Josh Berkus wrote: >> Thomas, >> >>> And don't forget to mention >>> PL/Java ;-) >> You'll be happy to know that the analyst had a specific question about java >> procedural support. >> >> BTW, what happened to the discussion of moving PL/java into core for 8.2? >> >> -- >> --Josh >> > Josh, > > We tried to include PL/java in our standard build, but the version of > the zlib library in the JVM is ancient and the incompatibility would > cause the DB to crash. Until we can get a JVM built against a current > library from Sun I think it would be a bad idea to have PL/java in > the core. > > Ken > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq >
Ken, I made some further investigation concerning this. I knew that Sun has addressed this bug long ago. Now I found two bug reports. Both bugs are marked as fixed/closed which make me curious. What Java version are you using when you encounter this problem? http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4953050 http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=5077725 I also managed to locate an old discussion around this that might be of interest: http://archives.postgresql.org/pgsql-hackers/2004-09/msg00715.php It indicates that the zlib is perhaps not needed at all in the backend. It's there because of "lazy" management of needed libraries. Regards, Thomas Hallgren Kenneth Marshall wrote: > On Thu, Feb 02, 2006 at 12:29:39PM -0800, Josh Berkus wrote: >> Thomas, >> >>> And don't forget to mention >>> PL/Java ;-) >> You'll be happy to know that the analyst had a specific question about java >> procedural support. >> >> BTW, what happened to the discussion of moving PL/java into core for 8.2? >> >> -- >> --Josh >> > Josh, > > We tried to include PL/java in our standard build, but the version of > the zlib library in the JVM is ancient and the incompatibility would > cause the DB to crash. Until we can get a JVM built against a current > library from Sun I think it would be a bad idea to have PL/java in > the core. > > Ken > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq >
Thomas Hallgren wrote: > I made some further investigation concerning this. I knew that Sun has > addressed this bug long ago. Now I found two bug reports. Both bugs are > marked as fixed/closed > which make me curious. What Java version are you using when you encounter > this problem? > > http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4953050 > http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=5077725 Bug 4953050 is marked to be fixed in "mustang(b65)", which is the next java release (1.6) not yet available. The bug has been reported against 1.4.2 and tiger-rc (1.5). The other bug has this note that seems to imply that this was an issue specific to Solaris and the fix might only apply to the specific platform. [...digs again...] Well, it seems this issue has been fixed in earlier versions, too. In a j2sdk 1.4.2_06 install I have here, doing "nm -D libzip.so | grep ' T '" shows that the exported symbols have prefixes now. (e.g. Java_java_util_zip_CRC32_update). Best Regards, Michael Paesold
Michael Paesold wrote: > Well, it seems this issue has been fixed in earlier versions, too. In a > j2sdk 1.4.2_06 install I have here, doing "nm -D libzip.so | grep ' T '" > shows that the exported symbols have prefixes now. (e.g. > Java_java_util_zip_CRC32_update). > Right. I made the same discovery. I also checked with a 1.5.0_05. Same thing there. Sun has back ported this to both the 1.4.2 and 1.5.0 branches. So this is likely to be a problem that's easy to fix. Regards, Thomas Hallgren
Thomas, Kenneth, > Bug 4953050 is marked to be fixed in "mustang(b65)", which is the next java > release (1.6) not yet available. The bug has been reported against 1.4.2 > and tiger-rc (1.5). > > The other bug has this note that seems to imply that this was an issue > specific to Solaris and the fix might only apply to the specific platform. Well, I have pull with both the Solaris team and Sun's Java people these days. If you can put the problem into a coherent memo, I can probably get it addressed in time for PostgreSQL 8.2, unless we run afoul of the JCP. --Josh
On Sun, 2006-02-05 at 02:25 +0100, Thomas Hallgren wrote:
> Question: When is zlib used in the backend today?
It isn't: zlib was only linked against the backend to simplify the build
process. In 8.2devel this has been corrected:
% ldd postgres
libcrypt.so.1 => /lib/libcrypt.so.1 (0x00002aaaaabc2000)
libresolv.so.2 => /lib/libresolv.so.2 (0x00002aaaaacf5000)
libnsl.so.1 => /lib/libnsl.so.1 (0x00002aaaaae0a000)
libdl.so.2 => /lib/libdl.so.2 (0x00002aaaaaf21000)
libm.so.6 => /lib/libm.so.6 (0x00002aaaab024000)
libc.so.6 => /lib/libc.so.6 (0x00002aaaab1a9000)
/lib64/ld-linux-x86-64.so.2 (0x00002aaaaaaab000)
See much discussion on this topic in the -hackers archive, and the fix
implemented by Martijn and Tom:
http://archives.postgresql.org/pgsql-patches/2005-11/msg00183.php
-Neil
Josh Berkus wrote: > Thomas, Kenneth, > > > Bug 4953050 is marked to be fixed in "mustang(b65)", which is the next java > > release (1.6) not yet available. The bug has been reported against 1.4.2 > > and tiger-rc (1.5). > > > > The other bug has this note that seems to imply that this was an issue > > specific to Solaris and the fix might only apply to the specific platform. > > Well, I have pull with both the Solaris team and Sun's Java people these days. > If you can put the problem into a coherent memo, I can probably get it > addressed in time for PostgreSQL 8.2, unless we run afoul of the JCP. So Sun is supporting PostgreSQL, but we have to write a memo to get a fix. This doesn't sound good. -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073
Bruce, > > Well, I have pull with both the Solaris team and Sun's Java people these > > days. If you can put the problem into a coherent memo, I can probably get > > it addressed in time for PostgreSQL 8.2, unless we run afoul of the JCP. > > So Sun is supporting PostgreSQL, but we have to write a memo to get a > fix. This doesn't sound good. By which I mean "a fairly coherent bug report." The Java people aren't going to do anything on the basis of me sending them a link to a mailing list archive. -- Josh Berkus Aglio Database Solutions San Francisco
Josh Berkus wrote: > Thomas, Kenneth, > >> Bug 4953050 is marked to be fixed in "mustang(b65)", which is the next java >> release (1.6) not yet available. The bug has been reported against 1.4.2 >> and tiger-rc (1.5). >> >> The other bug has this note that seems to imply that this was an issue >> specific to Solaris and the fix might only apply to the specific platform. > > Well, I have pull with both the Solaris team and Sun's Java people these days. > If you can put the problem into a coherent memo, I can probably get it > addressed in time for PostgreSQL 8.2, unless we run afoul of the JCP. > > --Josh > Thanks Josh, this is good to know the next time something pops up. No need for this particular issue though. I'd consider the case closed since a) Sun has already fixed this. The zlib fix is included in both the 1.4.2 and 1.5.0 branches at least 2 maintenance releases back and b) The PostgreSQL backend no longer link with zlib. Regards, Thomas Hallgren