fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged
hpt has joined #systemtap
hkshaw has quit [Ping timeout: 265 seconds]
hkshaw has joined #systemtap
hpt_ has joined #systemtap
hpt_ has quit [Client Quit]
<ericlee> Hi, how do I know the type of return value? like nfs.proc.open, it has 6 return values but how do I know what type shall I use when I want to print out the results?
ananth has joined #systemtap
ravi_ has joined #systemtap
hchiramm has quit [Quit: Leaving]
srikar_away is now known as srikar
ego_ has joined #systemtap
hchiramm has joined #systemtap
hchiramm has quit [Ping timeout: 250 seconds]
hchiramm has joined #systemtap
pwithnall has joined #systemtap
<lindi-> ericlee: 6 return values?
<ericlee> lindi-, yep
<lindi-> ericlee: it's a probe, it doesn't have a return value? or do you mean the return value of the nfs_open function?
<lindi-> ericlee: ah, I wouldn't call those return values
<ericlee> once it's triggered, 6 values are stored. Oh so what should be called?
<lindi-> ericlee: anyways, you need to read source of the tapset and linux to know the type
<lindi-> ericlee: I guess they are just values set by the probe
<lindi-> ericlee: nothing has returned yet
<ericlee> right, I read some and changed my code to run properly.
<ericlee> thanks.
irker223 has joined #systemtap
<irker223> systemtap: mcermak systemtap.git:refs/heads/master * release-3.0-217-g57aad8f / testsuite/systemtap.base/bz1269062.exp: Add testcase for RHBZ1269062. http://tinyurl.com/jq2afnc
mjw has joined #systemtap
hpt has quit [Quit: Lost terminal]
sjas_ has joined #systemtap
sjas has quit [Ping timeout: 250 seconds]
sjas_ is now known as sjas
scox has quit [Ping timeout: 260 seconds]
hkshaw has quit [Ping timeout: 240 seconds]
ananth has quit [Quit: Leaving]
ravi_ has quit [Remote host closed the connection]
hkshaw has joined #systemtap
tromey has joined #systemtap
Guest45280 has quit [Ping timeout: 264 seconds]
drsmith has joined #systemtap
wcohen has quit [Ping timeout: 250 seconds]
srikar is now known as srikar_away
ego_ has quit [Ping timeout: 244 seconds]
ananth has joined #systemtap
<irker223> systemtap: flu systemtap.git:refs/heads/master * release-3.0-218-g6585d28 / interactive.cxx: modify interactive mode load behavior http://tinyurl.com/zuuf9lg
scox has joined #systemtap
wcohen has joined #systemtap
brolley has joined #systemtap
<irker223> systemtap: flu systemtap.git:refs/heads/master * release-3.0-219-g97c16e1 / session.cxx: cleanup --help text http://tinyurl.com/jzlgxv2
flu has quit [Disconnected by services]
flu_ has joined #systemtap
<irker223> systemtap: flu systemtap.git:refs/heads/master * release-3.0-220-gfbd9e80 / man/stap.1.in: stap.1 man page interactive entry http://tinyurl.com/z6tvaxd
<fche> woohoo, he's on a roll
<csanting> hey fche, could you review some of my work?
<fche> surely
<csanting> thanks. mainly a new command for -i
<fche> consider using the ftw() api to traverse the sample directory rather than readdir
<fche> that way you get nested directories easily (should we ever need those)
<flu_> fche, currently the stap -i add command adds a function/probe even if p1/p2 fails, should those not be added at all to the script?
<fche> nah, a later edit command can let the user fix it
<fche> ie. yes add
<fche> and the user will fix it later
<flu_> ok
<csanting> fche, I'll update that command to use ftw, any other suggestions?
<fche> I was imagining something that searched the contents of the meta files
<fche> rather than just doing a directory listing
<fche> we have a keyword index (generated for readability)
<fche> we could extend the perl script that creates that to also make a machine-readable set
<fche> (stap -i sample> command readable)
<fche> almost like a little search index
<fche> ( by the way, + dirstr.append("/share/doc/systemtap/examples/") needs to be better generalized
<fche> maybe have auto{conf,make} pass ${docdir} to the c++ flags as an additional -Ddefine, kind of like STAP_PREFIX
<fche> 'cause share/doc/systemtap/examples is not universal. on older rhel, there are version numbers in there;
<fche> on other distros, share/doc might be doc/ only
<fche> e.g.
<fche> csanting, so e.g. imagine how you'd like to be able to find a sample for task FOO
<fche> where FOO is listed in the .meta file in a title/keywords/description/name type text field
ton31337 has quit [Ping timeout: 258 seconds]
<csanting> fche, ah, I see. Ok, I'll look into implementing it that way as well as updating the pathing, thanks for the review
ton31337 has joined #systemtap
<fche> csanting, yeah, how about sketching out some sample interactions (in a hypothetical console)
<fche> the user experience so to speak
<fche> and then work backward to code
<fche> it's the Agile Way (tm) :-)
<csanting> fche, good idea
ton31337 has quit [Ping timeout: 265 seconds]
hchiramm has quit [Ping timeout: 244 seconds]
<fche> here's something cute
<fche> cd ... testsuite/systemtap.examples
<fche> egrep '(name|title|description|keywords)' */*.meta | cut -f3- -d: | tr ' ' '
<fche> ' | grep . | tr A-Z a-z | tr -d '"().,-' | sort | uniq -c | sort -n
<fche> gives one a word list with the counts
<fche> so imagine having a table kind of sort of like that; stap> sample number ... would print the N meta files that use the word 'number' in those text fields
<fche> (that perl script could easily generate a word: path/to/file.meta index file
<fche> thence to let a user browse the meta file (as if cat $meta), implicitly loading the $meta.stp one as the current script
<fche> so it could add-to or overwrite the current script (probably overwrite makes more sense)
<fche> would that flow make sense?
ego_ has joined #systemtap
<csanting> fche, yes. Do you think it would be a good idea to "cleanup" that index to avoid cases with non-keywords?
* fche suggests trying it first without formal stopwords
<fche> maybe the user has a good reason for searching for 'it' :-)
<fche> and
<fche> anyways
<fche> another approach could be to just spawn out to a grep process
<fche> ie.,
<fche> stap> sample FOO => spawn("grep FOO $PATH_TO_SAMPLES/*/*/*.meta") and process the results
<fche> but anyway that's sort of goofy too. this functionality sounds simple enough to be done as a shell script, so no big deal how we actually do it
<csanting> fche, sounds good, I'll play around with these ideas
<fche> you da man
ton31337 has joined #systemtap
<irker223> systemtap: csanting systemtap.git:refs/heads/master * release-3.0-221-g2882325 / interactive.cxx: Update help commnd for interactive mode http://tinyurl.com/z8fnjk4
ton31337 has quit [Ping timeout: 244 seconds]
brolley has left #systemtap [#systemtap]
ton31337 has joined #systemtap
pwithnall has quit [Quit: pwithnall]
hchiramm has joined #systemtap
brolley has joined #systemtap
pfallenop has quit [Ping timeout: 258 seconds]
pfallenop has joined #systemtap
ego_ has quit [Ping timeout: 244 seconds]
<irker223> systemtap: csanting systemtap.git:refs/heads/master * release-3.0-222-g8099a52 / interactive.cxx: Interactive mode example commands http://tinyurl.com/zd9gfgk
<irker223> systemtap: flu systemtap.git:refs/heads/master * release-3.0-223-gcb57e97 / interactive.cxx: interactive mode command aliases http://tinyurl.com/hwjlvul
csanting has quit [Quit: Leaving]
flu_ has quit [Quit: flu_]
csanting has joined #systemtap
tromey has quit [Quit: ERC (IRC client for Emacs 25.1.3)]
mjw has quit [Quit: Leaving]
wcohen has quit [Remote host closed the connection]
scox has quit [Ping timeout: 250 seconds]
brolley has left #systemtap [#systemtap]
drsmith has left #systemtap [#systemtap]
wcohen has joined #systemtap