From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ulf Samuelsson Date: Mon, 07 Jul 2008 15:44:10 +0200 Subject: [Buildroot] svn commit: trunk/buildroot In-Reply-To: <20080707132821.GE15840@mx.loc> References: <20080704222950.E76B63C61F@busybox.net> <87fxqmg53j.fsf@macbook.be.48ers.dk> <48720878.3070602@atmel.com> <20080707123642.GA15840@mx.loc> <487212CC.7020303@atmel.com> <20080707131123.GC15840@mx.loc> <20080707132821.GE15840@mx.loc> Message-ID: <48721DAA.3080900@atmel.com> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Bernhard Fischer skrev: > On Mon, Jul 07, 2008 at 03:11:23PM +0200, Bernhard Fischer wrote: >> On Mon, Jul 07, 2008 at 02:57:48PM +0200, Ulf Samuelsson wrote: >>> -----BEGIN PGP SIGNED MESSAGE----- >>> Hash: SHA1 >>> >>> Bernhard Fischer skrev: >>>> On Mon, Jul 07, 2008 at 02:13:44PM +0200, Ulf Samuelsson wrote: >>>>> -----BEGIN PGP SIGNED MESSAGE----- >>>>> Hash: SHA1 >>>>> >>>>> Peter Korsgaard skrev: >>>>>> That breaks the source-check/external-deps check in >>>>>> Makefile.autotools.in (r22607). >>>>>> >>>>> Is not something like the following to be preferred? >>> Previously the special make targets to check availability >>> of packages on internet would download each target >>> even if the target existed in $(DL_DIR) creating >>> .tar.bz2., or it would not check >>> the package at all. >>> >>> Peter introduced a fix where he will treat the download >>> differently if the MAKELEVEL=0 (normal) or if the >>> MAKELEVEL=1 which is the case for make source-check >>> and make external-deps. >>> >>> When I introduced easy logging by "make log", >>> this breaks the autotools, since this also runs >>> at MAKELEVEL=1. >>> >>> I do not think that MAKELEVEL is a good variable >>> to detect if someone wants to do source-check >>> or external-deps. >> make source-check is _supposed_ to download all sources, regardless of > > Where download is not really appropriate since we do this (for speed and > volume reasons) deliberately with wget's --spider option, just as a > sidenote. > >> whether or not the package exists locally. Keep in mind that it is ment >> to check if packages and patches are (still) downloadable or not and >> explicitely disregarding any local stuff to simulate what users will >> experience. >> >> Perhaps the source-check target should consequently use a >> $(DL_DIR).check dir then. This would avoid breakage elsewhere. > > Due to spider, please disregard this comment about $(DL_DIR).check. I think the issue is that "source-check" and "external-deps" needs special treatment in autotools so that the WGET commands is executed regardless if the target package is present in DL_DIR or not. The WGET commands has been redefined to do WGET --spider or to run a script and neither of these will actually download the tarballs. What I question is the use of the MAKELEVEL variable to determine if this special treatment should be activated or not. >> Apart from that such a "log" target strikes me as unwarranted. >> Just >> $ (make) 2>&1 >& MY.log >> or >> $ nohup make & >> >> would a more sensible thing to do, IMHO. >> Not if you want your log file to have the same name as your project. Try typing "make > atngw100-expanded-3.log 2>&1" a couple of times, and you see what I mean. "make log" is so much easier, and you can use bash to just do "make log" by just pressing up-arrow . >> HTH, > _______________________________________________ > buildroot mailing list > buildroot at uclibc.org > http://busybox.net/mailman/listinfo/buildroot -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFIch2pAyRRH5cXxqwRArO1AKCyrz725gm471S/jZo+EABTwkZs9QCgmVMk IPBfdLBSIpA5EF57rOf1fBo= =NDo1 -----END PGP SIGNATURE-----