#emc-devel | Logs for 2008-11-16

[00:04:13] <BigJohnT> hmmm my frufru's came out nice except for one
[00:37:13] <CIA-38> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/config_gtk.c: update for modbus function read coils
[00:54:42] <KimK> There is another "thin" diagram on page 17 of the same manual. But I would like to praise the diagram on page 25, made by a different process, maybe the same LaTex(?) word processor as the rest of the manual? In any case, it's also a figure/diagram and it's quite clear. I saw some "fix me"s in the text, but I didn't see any other "thin" diagrams in the Developer's Manual.
[01:05:07] <KimK> I've got to shutdown for another drive swap, back in a bit
[08:38:43] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[16:42:11] <BigJohnT> I'm trying to get a makefile to build a html from a docbook file
[16:42:22] <BigJohnT> # Add your language file here:
[16:42:24] <BigJohnT> TARGETS = test2.html
[16:42:25] <BigJohnT> XSLTPROC = /usr/bin/xsltproc
[16:42:27] <BigJohnT> XSL = test2.xsl
[16:42:28] <BigJohnT> # xsltproc -o test2.html test2.xsl test2.xml
[16:42:30] <BigJohnT> %.html: %.xml $(XSL)
[16:42:31] <BigJohnT> @$(XSLTPROC) -o $@ $(XSL) $<
[16:42:33] <BigJohnT> all: $(TARGETS)
[16:42:34] <BigJohnT> clean:
[16:42:36] <BigJohnT> @rm -f *.html
[16:42:45] <BigJohnT> I get this error make: Nothing to be done for `all'.
[16:43:01] <BigJohnT> what did I miss?
[16:43:40] <SWPadnos> %.html: %.xml $(XSL)
[16:43:41] <seb_kuzminsky> BigJohnT: that means the all target is satisfied already, ie test2.html exists and is newer than the things it depends on (which look to be nothing)
[16:43:45] <SWPadnos> maybe that should be:
[16:43:48] <SWPadnos> %.html: %.xsl $(XSL)
[16:43:56] <SWPadnos> (note xSl vs xMl)
[16:45:28] <BigJohnT> yes I think your correct SWPadnos
[16:45:48] <SWPadnos> woohoo! (that rarely happens with makefile issues :) )
[16:45:54] <BigJohnT> ah well no...
[16:46:00] <SWPadnos> see!
[16:46:28] <SWPadnos> also you may have to remove the html file or touch the xsl file, since the target may be up to date
[16:46:29] <BigJohnT> maybe %.html: $(XSL) %.xml
[16:46:30] <jepler> no, I think you want the rule to say "to make an html file, take the corresponding xml file and the common xsl file"
[16:46:35] <seb_kuzminsky> makefiles are almost as bad as gcode
[16:46:35] <jepler> which is what it says
[16:46:36] <seb_kuzminsky> * seb_kuzminsky ducks
[16:46:46] <jepler> the xml file has to come first, so that $< gets the corresponding xml file
[16:47:25] <jepler> seb_kuzminsky: do you have a favorite make-alternative? I haven't found one I didn't hate even more than GNU make
[16:47:48] <seb_kuzminsky> me neither, though i havent looked around too much
[16:47:54] <BigJohnT> seems like seb_kuzminsky was the prize winner today
[16:47:58] <seb_kuzminsky> i've heard good things about scons
[16:48:13] <BigJohnT> I deleted the test2.html and ran make with sucess
[16:48:27] <jepler> seb_kuzminsky: I should probably take another look at scons
[16:50:25] <jepler> the problem with scons that I remember was that it had too many smart built-in rules; I wanted to replace a recursive-bsd-make, cross-compiling system with something new but all its built-in rules got in my way and I ended up looking elsewhere (non-recursive GNU make)
[16:51:39] <BigJohnT> jepler: so make takes the TARGETS and finds the .xml file by the same name and uses the common stylesheet to build each TARGET?
[16:51:41] <seb_kuzminsky> how did you like non-recursive make?
[16:52:35] <jepler> seb_kuzminsky: I'm a big believer in non-recursive make -- that's why we have a nearly-non-recursive make for emc2 (the kernel build system is a recursion)
[16:53:56] <SWPadnos> I'm a convert, based on the improvements I saw when jeff made the changes for EMC2
[16:54:10] <jepler> the makefiles still get messy over time, unfortunately
[16:54:22] <SWPadnos> I saw build times go down by 30%, and parallel makes always turn out right
[16:54:40] <seb_kuzminsky> was emc2 using autotools before jeff fixed it?
[16:55:15] <jepler> we never used automake as far as I remember
[16:55:18] <jepler> just hand-written recursive make
[16:55:31] <seb_kuzminsky> * seb_kuzminsky shudders
[16:57:57] <jepler> automake is just an impenetrable make system plus a terrible, terrible preprocessor to make up for shortcomings of non-gnu make
[16:58:14] <jepler> the gnu project would have been much better off saying "thou shalt use gnu make" than "thou shalt use automake"
[16:59:03] <seb_kuzminsky> jepler: that may be... I think of autotools as a (necessarily) complex language that compile to much more complex makefiles :-)
[17:01:21] <jepler> apparently it's possible to get a nonrecursive automake nowadays
[17:01:45] <jepler> https://forgesvn1.novell.com/svn/hula/branches/hula-store/hula/Makefile.am
[17:01:48] <jepler> e.g.
[17:01:51] <seb_kuzminsky> bbl
[17:07:18] <jepler> oh right, scons uses file hashes for dependency calculation, instead of timestamps
[17:07:54] <jepler> huh, it only takes 5x as long to md5sum all the source files in emc2 as to stat them (find ... -ls vs find ... -exec md5sum {} +)
[17:08:07] <jepler> I'm surprised the ratio is so low
[17:08:58] <jepler> hi LawrenceG -- someone on the mailing list was asking about your dspic servo amp. dunno if you caught that post..
[17:09:17] <jepler> From: "Kevin Pauba" <klpauba@gmail.com>
[17:09:20] <jepler> Subject: Re: [Emc-users] servo amplifier
[17:15:33] <LawrenceG> jepler, hi... no... just opening my eyes this morning... should be in the list to read this morning if it was on the mailing list
[17:19:31] <LawrenceG> gotit...
[17:56:56] <CIA-38> EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/common/machining_center.lyx: fix wandering figures
[18:25:01] <jepler> LawrenceG: I was wondering, what's the maximum quadrature rate from the servo motor?
[18:25:07] <jepler> for the dspic servo
[18:42:11] <BigJohnT> now I'm getting somewhere with docbook :)
[18:43:09] <alex_joni> cool
[18:43:17] <alex_joni> jepler: ever looked at ant?
[18:44:42] <jepler> alex_joni: only long enough to determine that it's java
[18:45:39] <jepler> oh and isn't it xml?
[19:13:19] <alex_joni> jepler: yeah, a bit ;)
[19:41:09] <LawrenceG> jepler, the motor goes to the hardware counter on the dspic... I have not been able to hit the limit yet. The PC side uses quadrature commands from the PC and is a little more limited, but supports pulse multiplying if required (each change from pc can be 1,2,...n counts of the motor encoder)
[19:43:25] <LawrenceG> jepler, all the servo internals work in motor encoder units and the position command from the PC just sets a position count that the loop tries to maintain
[20:11:35] <jepler> LawrenceG: ah, I didn't realize dspic had hardware quadrature counting -- that makes it about ideal for this
[20:19:03] <LawrenceG> jepler, it would be nice if it had 2 hardware encoder ports, then the PC side would be more robust as well.