Bug 186750
Summary: | Firefox with pango is very slow | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Johan Dahl <johan.dahl> |
Component: | firefox | Assignee: | Christopher Aillon <caillon> |
Status: | CLOSED INSUFFICIENT_DATA | QA Contact: | |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 5 | CC: | dgunchev, jake, johan.dahl, jws, mcepl, moneta.mace, wes, wtogami |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2007-05-21 14:53:21 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Johan Dahl
2006-03-26 02:06:04 UTC
I have a similar system (Athlon XP 1900+) and have this same problem. Firefox is more responsive if I use the version from mozilla.org, which I'm guessing does not have pango, nor all of the language extensions (which should be another bug report in itself). Hopefully this will be looked into even though it's not a huge bug, because it makes Firefox seem really bloated. Epiphany uses the same rendering engine as far as I know and it seems to load pages decently faster than the built-in Firefox. I can confirm that firefox + pango is very slow on my system, fustratingly so. I was getting very agro about it untill I realised what had happened. I was using fc4 with the downloaded version firefox 1.5 from mozilla.org then upgraded to fc5 with its firefox + pango enabled version and the system was much slower. I have a dell fpw 2405 @ 1920x1200 + nvidia 7800 and I think the size of the screen may exaserbate the problem. Fedora will not turn off pango by default. Instead of complaining, I suggest that you help development of pango, because pango will become the default rendering engine for upstream's Mozilla in the near future. I observe varied behaviour depending, it seems, on the history of the system. An up-to-date FC4 upgraded to FC5 has the problem and it makes the browser very hard to use. A similar set of hardware, where the main difference is 1GB dual channel v.s. 512MB single channel memory, has no observable ill-effects. These are 2/2.2GHz K8s. It looks like setting MOZ_DISABLE_PANGO=1 in the environment before starting firefox addresses the performance issue, until pango's performance can be improved. Furthermore this MOZ_DISABLE_PANGO=1 setting is written in the releases notes, and the performance issue is written as known. This bug should be closed. Not a bug? It's been a little while since there's been a comment here. Not trolling, but I came to file a suggestion to disable Pango by default (and leave it to the release notes or an overriding package to re-enable Pango so that people can take advantage of it for specific environments). Before I go off and suggest something stupid, though: reading through this report, I'm seeing "Instead of complaining, I suggest that you help development of pango, because pango will become the default rendering engine for upstream's Mozilla in the near future." So, sorry to bug people, but could someone help me understand better: does this mean that the Mozilla dev team is phasing out the "non-Pango" rendering engine in favor of Pango, and and Pango would be code with the most development moving forward? Because, if so, leaving Pango on in the default scripts seems the better thing to do. (Hell, if that's the way things go, I may have to look at the Mozilla base code, which I know I haven't done in years.) If Pango is just an option, though: it's been about ten months since the statement was made, and about eight since the last comment on this. So, I not knowing the status, I can only comment on the effect that it has on our public image as a community: to an beginner or outsider, it just translates to "Fedora is slow" or "Fedora crashes." And, yes, I've actually heard this comment from people that I've installed systems for (and I've forgotten to change the scripts.) And, I'm sorry to say (because I don't want to raise any hackles, and I know the Pango dev team is probably working very hard on the issue), we should remember that this problem is observed in two very high-profile applications to nontechnical users: the Web browser and the email client. From a personal perspective, I can say that I run a 1920x1200 resolution, which definintely exacerbates the problem. At that resolution, both Thunderbird and Firefox have actually crashed and/or become unresponsive during font resize (I can't recall anything so extreme in rendering). Now, that's all right by me: to get through the day, I know how to go in and modify the startup scripts. And, since I actually do things like read release notes every time they come out and read through bug report conversations; well, it really doesn't effect me at all. The only thing concerning me is the public face of Fedora. Again, if the non-Pango engine is being phased out of development in the Mozilla community; well, Fedora is here to forge forward, and none of this matters. However, if that's not the case, it'd probably be a good thing to bring up changing the default scripts, and allow the Pango developers some time to continue their work on improving performance of the engine. Can anyone enlighten me as to where Pango "fits in" to the plan? And wow, please excuse all the typos in that. Damn. Upstream Firefox is moving toward Pango by default. Pango is already needed by default in order to support many global languages. Fedora will not disable it by default and make these users turn it on manually with an ugly hack. All this is fixed in FC6, though, and pango is enabled by default there. Just to be sure -- could you, Johan and Jake, confirm that you can reproduce slowness and crashes with default Pango setting (i.e., please, use standard starting script) in the current FC6? No response from reporters, closing as INSUFFICIENT_DATA. If anyone has some information on this bug, please, reopen with additional data. |