228113 2021-07-20 11:22:32 -0700 Add .gitignore file for CISupport 2021-07-28 09:55:02 -0700 1 1 1 Unclassified WebKit Tools / Tests WebKit Nightly Build Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 1 aakash_jain aakash_jain aakash_jain ap dewei_zhu jbedard ryanhaddad webkit-bug-importer oldest_to_newest 1778211 0 aakash_jain 2021-07-20 11:22:32 -0700 Add .gitignore file for CISupport, especially for various buildbot related services we host from that directory. 1778212 1 433884 aakash_jain 2021-07-20 11:23:47 -0700 Created attachment 433884 [fast-cq] Patch 1778226 2 ap 2021-07-20 11:54:49 -0700 Is it possible to NOT have these files in a checkout? This seems like papering over the issue. 1778227 3 aakash_jain 2021-07-20 11:59:54 -0700 These log files are generated on the server hosting the applications (not on individual user's machines). They pollute the 'git status' command on those servers. Those log files are part of the intended functionality (e.g.: twistd.log for buildbot) 1778230 4 ap 2021-07-20 12:04:51 -0700 I understand, it's just unclear to me why they have to be here even on the server. It seems like it can only add trouble because of potentially conflicting access restrictions, and/or be bad for security to mix source code and runtime data. 1779736 5 jbedard 2021-07-27 07:12:23 -0700 (In reply to Alexey Proskuryakov from comment #4) > I understand, it's just unclear to me why they have to be here even on the > server. It seems like it can only add trouble because of potentially > conflicting access restrictions, and/or be bad for security to mix source > code and runtime data. If this is the concern, I think we need to change how we have buildbot services running. Given that the current instructions for running our buildbot instances (both locally and on the production servers) has them running out of Tools/CISupport/<host>, this is the right change. Even if we change where we're running our production servers, this is still applicable for local instances, and our other services (even those cloud hosted) tend to run "in place" 1779797 6 webkit-bug-importer 2021-07-27 11:23:24 -0700 <rdar://problem/81172638> 1779900 7 ews-feeder 2021-07-27 16:59:10 -0700 Committed r280362 (240009@main): <https://commits.webkit.org/240009@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 433884. 1779919 8 ap 2021-07-27 17:39:53 -0700 > If this is the concern, I think we need to change how we have buildbot > services running. Given that the current instructions for running our > buildbot instances (both locally and on the production servers) has them > running out of Tools/CISupport/<host>, this is the right change. Even if we > change where we're running our production servers, this is still applicable > for local instances, and our other services (even those cloud hosted) tend > to run "in place" I don't think that you understood my comment. Running from a checkout is completely orthogonal to having write access to the checkout and messing it up! Neither local nor cloud nor any other deployments have to write to write to the directory where the code is in. This is like saying that Apache logs must be in /usr/sbin, because that's where the httpd binary is. 1780059 9 jbedard 2021-07-28 08:07:18 -0700 (In reply to Alexey Proskuryakov from comment #8) > > If this is the concern, I think we need to change how we have buildbot > > services running. Given that the current instructions for running our > > buildbot instances (both locally and on the production servers) has them > > running out of Tools/CISupport/<host>, this is the right change. Even if we > > change where we're running our production servers, this is still applicable > > for local instances, and our other services (even those cloud hosted) tend > > to run "in place" > > I don't think that you understood my comment. > > Running from a checkout is completely orthogonal to having write access to > the checkout and messing it up! Neither local nor cloud nor any other > deployments have to write to write to the directory where the code is in. > > This is like saying that Apache logs must be in /usr/sbin, because that's > where the httpd binary is. Ah, I see the point. Seems like we would want a "log" directory inside the checkout, in that case. Sort of like what we do for autoinstall and WebKitBuild? 1780086 10 ap 2021-07-28 09:55:02 -0700 I'd say somewhere in /var/log, not inside the checkout? Perhaps I'm missing something too. 433884 2021-07-20 11:23:47 -0700 2021-07-27 16:59:12 -0700 [fast-cq] Patch bug-228113-20210720142346.patch text/plain 867 aakash_jain SW5kZXg6IFRvb2xzL0NoYW5nZUxvZwo9PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09Ci0tLSBUb29scy9DaGFuZ2VMb2cJKHJl dmlzaW9uIDI4MDA4OSkKKysrIFRvb2xzL0NoYW5nZUxvZwkod29ya2luZyBjb3B5KQpAQCAtMSwz ICsxLDEyIEBACisyMDIxLTA3LTIwICBBYWthc2ggSmFpbiAgPGFha2FzaF9qYWluQGFwcGxlLmNv bT4KKworICAgICAgICBBZGQgLmdpdGlnbm9yZSBmaWxlIGZvciBDSVN1cHBvcnQKKyAgICAgICAg aHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcuY2dpP2lkPTIyODExMworCisgICAgICAg IFJldmlld2VkIGJ5IE5PQk9EWSAoT09QUyEpLgorCisgICAgICAgICogQ0lTdXBwb3J0Ly5naXRp Z25vcmU6IEFkZGVkLgorCiAyMDIxLTA3LTIwICBTYW0gU25lZGRvbiAgPGdzbmVkZGVyc0BhcHBs ZS5jb20+CiAKICAgICAgICAgQ2hhbmdlIHJlZmVycmVyLXBvbGljeSBkZWZhdWx0IHRvIHN0cmlj dC1vcmlnaW4td2hlbi1jcm9zcy1vcmlnaW4KSW5kZXg6IFRvb2xzL0NJU3VwcG9ydC8uZ2l0aWdu b3JlCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT0KLS0tIFRvb2xzL0NJU3VwcG9ydC8uZ2l0aWdub3JlCShub25leGlzdGVu dCkKKysrIFRvb2xzL0NJU3VwcG9ydC8uZ2l0aWdub3JlCSh3b3JraW5nIGNvcHkpCkBAIC0wLDAg KzEsOSBAQAoraHR0cC5sb2cqCitld3MubG9nKgorZXdzX2NvbnNvbGUubG9nKgordHdpc3RkLmxv ZyoKK3R3aXN0ZC5waWQKK3N0YXRlLnNxbGl0ZQorcGFzc3dvcmRzLmpzb24KK3B1YmxpY19odG1s Lword29ya2Vycy8K