Re: [gccsdk] RO support being removed from curl
On Thu, May 04, 2017 at 03:17:16PM +0100, Ralph Corderoy wrote: > Hi Theo, > > > Unfortunately the commit in curl upstream that introduced it has no > > indication of who originated it. so we're still in the dark. I > > haven't dug through the curl mailing list history though. > > I had a bit of a go at that earlier, both manually skimming two of the > list archives backwards from the time of the commit, and getting Google > to search that site for "RISC OS" or "USE_ENVIRONMENT", etc. Didn't > turn anything up. Others may do better. :-) I had a bit more of a dig and came up with: https://curl.haxx.se/mail/archive-2002-03/0145.html which says the relevant port was from Michael Curtis. That eventually led me to: http://web.archive.org/web/20040205143254/http://homepages.ihug.co.nz:80/~jamjars/michael/software.html which has a download... that doesn't have a frontend. It also doesn't seem to have the --environment option available. So my guess is that this isn't actually used for anything, and is safe to remove. The other patches from Michael Curtis seem to be related to building natively on RISC OS (ie without autoconf) - while that might be useful for some, I have no idea if the GCC from today will build it with the same flags as the ones committed 15 years ago. I think in that case the only sustainable route is cross-compilation with autoconf, unless someone wants to maintain curl's upstream RISC OS-native config.h Theo ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
Re: [gccsdk] RO support being removed from curl
On Thu, May 04, 2017 at 02:07:25PM +0100, Lee Noar wrote: > On 04/05/17 11:11, Theo Markettos wrote: > >I suspect losing it won't actually affect RISC OS beyond whatever GUI it was > >intended for, but curious to know what that was. > > If necessary (not that I know anything about curl), we could just add > back the missing bits in the autobuilder. We could, but upstream say it's a pain to maintain because it gets in the way of other things. So it's probably not something we want to maintain because I sense upstream is going to put something in that breaks a patch. If the maintainer of the GUI that uses it steps forward and says they're willing to keep a patch up to date, then we could build with that patch I suppose. Unfortunately the commit in curl upstream that introduced it has no indication of who originated it. so we're still in the dark. I haven't dug through the curl mailing list history though. Theo ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
Re: [gccsdk] Building GCCSDK with modern GCC + makeinfo
On 24/04/17 21:51, Jeffrey Lee wrote: Hi, Over the weekend I had a go at building GCCSDK on a fresh install of Lubuntu 17.04, using the default GCC + makeinfo/texinfo packages (GCC 6.3.0 & makeinfo 6.3). GCC 6.3 didn't like the version of PPL we're using (it complained about some classes which contained a "[]" flexible array as the last member). So I've fixed that by updating to the latest version of PPL (which allows us to drop some now redundant patches, but also introduces a couple of new patches because GCC and cloog had bogus PPL version checks). AIUI more recent versions of GCC have dropped PPL in favour of a different library ("isl"), so I'll leave it to you to decide whether updating the PPL version is worth it or not (it's possible it's affected code generation/optimisation, since GCC built with new PPL seems to produce different binaries compared to other builds of GCC with old PPL) Meanwhile, makeinfo barfed at a couple of pieces of dodgy syntax in the GCC docs. Checking against the latest GCC sources shows that they've changed that part of the docs to use an alternative syntax, which looks like it should be supported under all the makeinfo versions we support: https://gcc.gnu.org/viewcvs/gcc/trunk/gcc/doc/gcc.texi?r1=203064=203065; After patching the above issues I was able to build functioning versions of the cross-compiler and native compiler. Patches attached. Thanks for that Jeffrey, I'll probably update PPL to begin with and look at ISL later. Lee. ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
Re: [gccsdk] RO support being removed from curl
On 04/05/17 11:11, Theo Markettos wrote: I suspect losing it won't actually affect RISC OS beyond whatever GUI it was intended for, but curious to know what that was. If necessary (not that I know anything about curl), we could just add back the missing bits in the autobuilder. Lee. ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
Re: [gccsdk] RO support being removed from curl
On Thu, May 04, 2017 at 10:44:54AM +0100, Ralph Corderoy wrote: > Hi, > > Jeremy wrote: > > The pull-request for doing so is already submitted, see link below. > > The email is https://curl.haxx.se/mail/archive-2017-05/0004.html and the > PR is https://github.com/curl/curl/pull/1463 with > https://github.com/curl/curl/pull/1463/commits/85b3072d422d0c25cb8c5d724206bebcc222f149#diff-10de2e0e2a5a0a4f65bb973cfd304c5aL34 > being the interesting bit listing the environment variables. We build command-line curl in the GCCSDK autobuilder - we consume sources from Debian, to save having to have interactions with hundreds of upstream projects which Debian already do. That said, --environment seems to set system variables: - {"curl_url_effective", CURLINFO_EFFECTIVE_URL, writeenv_STRING}, - {"curl_http_code", CURLINFO_RESPONSE_CODE, writeenv_LONG}, - {"curl_time_total", CURLINFO_TOTAL_TIME, writeenv_DOUBLE}, - {"curl_time_namelookup", CURLINFO_NAMELOOKUP_TIME, writeenv_DOUBLE}, - {"curl_time_connect", CURLINFO_CONNECT_TIME, writeenv_DOUBLE}, - {"curl_time_pretransfer", CURLINFO_PRETRANSFER_TIME, writeenv_DOUBLE}, - {"curl_time_starttransfer", CURLINFO_STARTTRANSFER_TIME, - writeenv_DOUBLE}, - {"curl_size_header", CURLINFO_HEADER_SIZE, writeenv_LONG}, - {"curl_size_request", CURLINFO_REQUEST_SIZE, writeenv_LONG}, - {"curl_size_download", CURLINFO_SIZE_DOWNLOAD, writeenv_DOUBLE}, - {"curl_size_upload", CURLINFO_SIZE_UPLOAD, writeenv_DOUBLE}, - {"curl_speed_download", CURLINFO_SPEED_DOWNLOAD, writeenv_DOUBLE}, - {"curl_speed_upload", CURLINFO_SPEED_UPLOAD, writeenv_DOUBLE}, Does anybody use these? I'm imagining they're probably intended for a progress GUI driving the curl binary (rather than using libcurl, which has scheduling complications). The only frontend I could find is this one: http://www.strcprstskrzkrk.co.uk/curl which looks like it's using the Frontend module - ie there's no GUI beyond setting the options to run the command. I suspect losing it won't actually affect RISC OS beyond whatever GUI it was intended for, but curious to know what that was. Theo ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
Re: [gccsdk] RO support being removed from curl
Hi, Jeremy wrote: > The pull-request for doing so is already submitted, see link below. The email is https://curl.haxx.se/mail/archive-2017-05/0004.html and the PR is https://github.com/curl/curl/pull/1463 with https://github.com/curl/curl/pull/1463/commits/85b3072d422d0c25cb8c5d724206bebcc222f149#diff-10de2e0e2a5a0a4f65bb973cfd304c5aL34 being the interesting bit listing the environment variables. -- Cheers, Ralph. https://plus.google.com/+RalphCorderoy ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK
[gccsdk] RO support being removed from curl
I read, by accident last night, a post on the curl-users mailing list, a suggestion by its maintainer that RO-specific support should be removed from the source. He said: The --environment command line option has existed since April 2002 and is documented to be specifically for RISC OS builds. It is also enabled in src/makefile.dj for DOS-builds. I now want to remove that option and all code related to it since it is such a niche feature, there doesn't seem to be any RISC OS builds updated since years and I see no reason why DOS builds specifically should need/use this either. The pull-request for doing so is already submitted, see link below. If you object to me merging this, please speak up and tell me why! I haven't a clue whether that would cause any current RO users a problem. As the maintainer mentioned this in a post entitled: --environment anyone? I've commented that no RO user of curl would necessarily realise that it meant them... I also suggested that he contact ROOL to see if they had a view on this, but I'm not sure that he is going to. So I did... and Ben Avison replied suggesting I mention it here. The curl mail list can be found at: https://curl.haxx.se/mail/list.cgi?list=curl-users (It's been ages since I used RO, but I know it's still in use.) -- Jeremy Nicoll - my opinions are my own ___ GCCSDK mailing list gcc@gccsdk.riscos.info Bugzilla: http://www.riscos.info/bugzilla/index.cgi List Info: http://www.riscos.info/mailman/listinfo/gcc Main Page: http://www.riscos.info/index.php/GCCSDK